The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
/cpnomic
/Review
January/February 1995
Volume 80, Number 1
r\
Id) ü g g 11 f E. |j
M
Federal Reserve
Bank of Atlanta
In This Issue:
5ome Lessons from Basic Finance for
Effective Socially Responsible Investing
inflation and Inflation Forecasting:
An Introduction
Iteview Essay—An Economist's
Perspective o n History
2 2 «
mJ
FR3 RESEARCH LIBRARY
V.
.
/
•
/
K
\
-
•
'
•
-
-
•-/.-
/
-
-
-
. - .
<
- V-
'
J
.
y •
•
' '
J- y
•• ) I
- ~ r
V-r
- v
>
Y
- -V
r
r
•-- •'
•
.
-y
\
'
•
.
~
/ /v-
-
_ k
. v
: -
•
.
1
1
r,,
'
j' •
c
~
"•
y
J
•
<
.
.
-
'
•
- V v , -'
• '
v-
\
jr-r- M
•xi
/
, Á
\
^
.
•
• • is-
'
;
Y
y
- . .
\
v
-> •
• —'
- '•i-- .
• ...
%
* ,
•
T
V<
•
- -
;
N,-
'
• •- '
--
w
' -
-
• >
(
'••
-
^
"
• /
¿
• K
' ••'- -,
.
>
..-
f.
I
'
-
»•>
-
. •
•
r
V
"N • "
•
•' N . •
.Y
,
J • •
'
.
-w >
—
•
Y
;
'
•
.
'
t
,
- /
V"
-
-
-
:
•
- ñ.
• ^
'
•
;
v y
1
' j .
-
^
-
,\
•y ' •
•.
I
>
y y
y - y --s?
x , r y . yv. ^
•
'
y
• /'•'•
>' - :
http://fraser.stlouisfed.org/
;
Federal
Reserve
Bank of St. Louis
'•
-y
i.
' y
''v
•e • •
.. .
y
1
/
*"•,
-
•
- :
i
. •-:
•
,—- '
r
i -
— . -V -,
V
;
•
'
. : .•
-. -
- • ..
-
•
"
y->
••> •
..... -
^ _
x'
<
' V. ,
y y
-'
-->
-
«'f
/
V"
.
,
-
1
'
. .
...
1
• ' • f .,..
.
< ,
v'
^
•
.
y.
:
• .
(
• f
v
L,
January/February 1995, Volume 80, Number 1
Federal Reserve
Bank of Atlanta
President
Robert P. Forrestal
Senior Vice President and
Director of Research
Sheila L. Tschinkel
Research Department
B. Frank King, Vice President and Associate Director of Research
William Curt Hunter, Vice President, Basic Research and Financial
Mary Susan Rosenbaum, Vice President, Macropolicy
Thomas J. Cunningham, Research Officer, Regional
Eric M. Leeper, Research Officer, Macropolicy
William Roberds, Research Officer, Macropolicy
Larry D. Wall, Research Officer, Financial
Public Affairs
Bobbie H. McCrackin, Vice President
Joycelyn Trigg Woolfolk, Editor
Lynn H. Foley, Managing Editor
Carole L. Starkey, Graphics
Ellen Arth, Circulation
M
H
^
The Economic Review of the Federal Reserve Bank of Atlanta presents analysis of economic
and financial topics relevant to Federal Reserve policy. In a format accessible to the nonspecialist, the publication reflects the work of the Research Department. It is edited, designed, produced, and distributed through the Public Affairs Department.
Views expressed in the Economic Review are. not necessarily those of this Bank or of the Federal Reserve System.
Material may be reprinted or abstracted if the Review and author are credited. Please provide the
Bank's Public Affairs Department with a copy of any publication containing reprinted material.
Free subscriptions and limited additional copies are available from the Public Affairs Department, Federal Reserve Bank of Atlanta, 104 Marietta Street, N.W., Atlanta, Georgia 30303-2713
(404/521-8020). Change-of-address notices and subscription cancellations should be sent directly to the Public Affairs Department. Please include the current mailing label as well as any new
information. ISSN 0732-1813
Federal Reserve Bank of Atlanta Economic Review
January/February 1995, Volume 80, Number 1
Some Lessons from Basic
Finance for Effective
Socially Responsible
Investing
Larry D. Wall
Given the vast sums of money that mutual and pension fund
managers invest, an important question is how they should go
about deciding which assets, especially which stocks, they should
purchase. One school of thought argues that investment policies
should reflect some set of social values. This study examines
three questions about the financial implications of effective socially responsible investing in common stocks—that is, socially
responsible investment intended to change firms' behavior.
The first question concerns what socially responsible investors
can do to effectively influence firms' investment policies. The second question is, under what conditions, if any, will the securities
markets permit effective socially responsible investment? Third,
what impact will socially responsible investment have on the performance of portfolios that follow it?
The analysis has two implications for fund managers and
investors who want to change firms' behavior. The first is that
the investment strategy should focus on buying shares of small
socially responsive firms. The second is that investors who owned
targeted socially responsible stocks before socially responsible investment began will realize above-market rates of return in the
short run; however, once socially responsible investors stop bidding up the price, investors will receive reduced returns.
-/inflation and
Inflation Forecasting:
An Introduction
Ellis W. Tallman
Although current inflation rates are relatively benign, the costs
of unexpected inflation, even at low rates, remain substantial for
individual firms and consumers. Many types of planning decisions, such as businesses' and governments' plans for expected
expenses and revenues, hinge on inflation forecasts.
This article provides an overview of the effects of inflation and
the significance of inflation forecasting. The author first considers
how forecasting models are specifically designed to fill the needs
of particular users. The analysis examines two statistical models—the Phillips curve and money demand/monetarist models—
that employ standard economic theory to suggest variables that
help predict inflation. Forecasts from a simple, but widely used,
version of each model are then compared with simple time series
models that include only past data on inflation. This comparison
using standard accuracy criteria shows that the economic models
did not perform much better than the simplest time series forecasting model. The author concludes that future research should focus
on estimating dynamic models that are by design more structural
and that may help uncover the sources of inflation.
/feview Essay—An
Economist's Perspective
o n History: Institutions,
Institutional Change, and
Economic
Performance
Andrew C. Krikelas
Douglass North earned a share of the 1993 Nobel prize for economics for two decades of research that culminated in the development of an innovative framework for analyzing economic history.
This review essay discusses the book that most comprehensively
presents North's paradigm, which characterizes history as the
record of an evolving game in which institutions, organizations,
and individuals function as the rules, teams, and players. Through
examples, the reviewer illustrates how North's game theoretic
paradigm can serve not only as a tool for analyzing historical
events but also as a methodological bridge between the diverse
branches of the social sciences and humanities.
Some Lessons from
Basic Finance for Effective
Socially Responsible Investing
Larry D. Wall
7u
/ I
/
I
/
A
The author is a research
officer in the financial
section
of the Atlanta Fed's
research
department. He thanks Peter
Ahken, Andy
Krikelas,
and Pam Peterson for
helpful
comments.
Federal
Reserve Bank of Atlanta
iM
/ M
/
m
m
utual f u n d s and pension plans are t w o of the most rapidly
g r o w i n g types of financial intermediaries in the U n i t e d
States. Together they accounted for almost 4 0 percent of
all financial intermediaries' assets at the end of 1993 (see
V
M
George G. K a u f m a n and Larry R. M o t e 1994). Given the
vast sums of m o n e y that mutual and pension f u n d m a n a g e r s invest, an important question is h o w they should g o about deciding which assets, especially w h i c h stocks, they should p u r c h a s e . F u n d s that invest in s t o c k s h a v e
generally been actively m a n a g e d — t h a t is, the f u n d m a n a g e r buys stocks he
or she believes are undervalued and sells stocks thought to be overvalued.
This approach to investing has been challenged in recent years, however.
M o d e r n finance theory presents one challenge, arguing that f u n d s should
buy widely diversified portfolios w h o s e composition roughly approximates
the entire stock market. This position is based on the efficient markets argum e n t , which holds that stocks are correctly priced and investors cannot syst e m a t i c a l l y f i n d s t o c k s that are e i t h e r u n d e r - o r o v e r v a l u e d . F r o m this
perspective, active m a n a g e m e n t of stocks only wastes value by increasing
transactions costs. 1
T h e other challenge to traditional active m a n a g e m e n t c o m e s f r o m those
w h o argue that investment policies should reflect s o m e set of social values. 2
A d v o c a t e s of this point of view argue that socially responsible investing
m a y help improve the world. S o m e would also argue that socially responsive c o r p o r a t i o n s m a y be m o r e p r o f i t a b l e in the l o n g run b e c a u s e their
Economic Review 9
workers are likely to be m o r e productive and because
changing government regulations and social pressures
are less likely to affect such corporations adversely.
I n v e s t m e n t b a s e d on social g o a l s falls u n d e r the
b r o a d c a t e g o r y of socially r e s p o n s i b l e i n v e s t m e n t . 3
Given that different people place different priorities on
various social goals, there is no single, universally accepted definition of social responsibility. In practice,
socially responsible investment criteria work by limiting the universe of stocks that a f u n d m a n a g e r considers via s o m e combination of negative (exclusionary)
and positive (inclusionary) screens. For example, participation in the tobacco and w e a p o n s industries might
constitute "socially undesirable" activities included in
n e g a t i v e s c r e e n s ; " s o c i a l l y d e s i r a b l e " a c t i v i t i e s allowed under positive screens might include following
g o o d e n v i r o n m e n t a l p r a c t i c e s and s e e k i n g to h a v e
good employee relations.
While the economic merits of active versus passive
m a n a g e m e n t of funds have been a source of continuing
d e b a t e a m o n g a c a d e m i c s and practitioners, the econ o m i c implications of socially responsible investing
have received somewhat less attention. It is likely that
concerning this issue the gap between academic opinions and practitioner v i e w s is s o m e w h a t larger. T h i s
study seeks to close that gap by focusing on what will
here be called effective socially responsible investment
in c o m m o n stocks—that is, the buying (selling) of publicly traded c o m m o n stocks on the basis of s o m e social
criteria in order to increase (decrease) the f i r m s ' level of
investment in plant and equipment. In other words, the
discussion analyzes the implications of socially responsible investment policies that a mutual f u n d , pension
fund, or individual investor might follow. In particular,
this study addresses three questions: (1) W h a t must socially responsible investors d o to effectively influence
f i r m s ' investment policies? (2) Under what conditions,
if any, will securities markets permit effective socially
r e s p o n s i b l e i n v e s t m e n t ? (3) If e f f e c t i v e socially responsible investment is possible for at least s o m e securities, h o w will i n c l u d i n g s u c h s e c u r i t i e s a f f e c t the
performance of portfolios? The analysis is largely based
on fundamental finance principals covered in standard
undergraduate finance texts; in particular this study will
use the terminology and notation of P a m e l a P. Peterson's (1994) introductory text. 4
The results of this study will be most relevant to
i n v e s t o r s with social c o n c e r n s w h o w a n t to c h a n g e
f i r m s ' b e h a v i o r . T h i s d i s c u s s i o n will i n d i c a t e b o t h
w h e n s i g n i f i c a n t social c h a n g e s are likely to result
from investment decisions and what the implications
of effective socially responsible investment will be for
2
Econom ic Review
an i n v e s t o r ' s rate of return. T h e analysis will be f a r
less valuable f o r investors w h o s e primary concern is
m a t c h i n g their i n v e s t m e n t s to their personal v a l u e s
r a t h e r than s e r v i n g larger social g o a l s . E x i s t i n g finance theory, which generally a s s u m e s that investors'
only c o n c e r n is the risk and return f r o m their asset
h o l d i n g s , h a s little to say a b o u t o p t i m a l i n v e s t m e n t
policies for investors w h o also want to take into account the w a y in w h i c h f i r m s obtain their e a r n i n g s .
The analysis will also h a v e limited value for investors
w h o plan on b u y i n g securities that are not publicly
traded and w h o t h e r e f o r e f a c e issues very d i f f e r e n t
f r o m those discussed below."1
.Existing Studies of Socially
Responsible Investment
Efficient M a r k e t Analysis. Financial e c o n o m i s t s
usually begin any analysis of stock markets with the eff i c i e n t m a r k e t s h y p o t h e s i s , w h i c h h o l d s that s t o c k
prices reflect all publicly available information. 6 In this
context of what is called "semistrong efficiency," n o
investor may use publicly available information to obtain an above-market rate of return without a c o m m e n surate increase in risk. 7 Socially responsible investment
portfolios m u s t earn a market rate of return after adjusting for their level of (nondiversifiable) risk.
A r g u m e n t s for Socially Responsible Investment.
Advocates of socially responsible investment attack the
efficient markets perspective in t w o ways. First, they
argue that a large and growing body of evidence suggests that markets are not semistrong efficient. Maria
O ' B r i e n Hylton (1992), for example, points to a n u m ber of studies that suggest inefficiency in financial markets. A c c o r d i n g to H y l t o n t h e s e s t u d i e s i m p l y that
socially responsible investors could profit f r o m market
inefficiency resulting from speculation.
Second, some advocates of socially responsible investment argue that, given the inefficiency of the market,
"socially responsible" firms may generate above-market
returns and "socially irresponsible" firms may generate
below-market returns (for example, see Severyn T. Bruyn
1987). For instance, firms that follow socially irresponsible environmental policies may face increased costs and
reduced market share as governments and c o n s u m e r s
hold corporations to ever-increasing standards of environmental cleanliness. In contrast, firms that follow environmentally friendly policies may have increasing
market opportunities around the world as people become
more concerned about their environment.
January/February 1995
C o u n t e r a r g u m e n t s . Although the claim that markets have been f o u n d to be inefficient has substantial
s u p p o r t , it is not s t r o n g e n o u g h to p r o v i d e a c o m pelling argument in support of socially responsible investment. N u m e r o u s studies have found evidence that
markets are in fact efficient in a w i d e variety of circ u m s t a n c e s . 8 It is t h e r e f o r e a m i s t a k e to g e n e r a l i z e
f r o m the findings of the f e w studies that suggest market i n e f f i c i e n c y to c o n c l u d e that m a r k e t p r i c e s are
consistently inaccurate indicators of a f i r m ' s value.
Moreover, the case for socially responsible investment would not be proved even if it were shown to be
true both that (1) financial markets are inefficient and
(2) socially responsible firms are m o r e profitable than
irresponsible firms. 9 The source of the market inefficiency could be unrelated to whether f i r m s are socially
responsible, in which case an investor f o l l o w i n g socially responsible investment policies would not earn a
higher rate of return. A n o t h e r possibility is that firms
that generate high profits are m o r e likely to be overvalued by the stock market in which investors following socially responsible investing m a y earn below
market rates of return.
Empirical Evidence. T h e theoretical arguments
that socially responsible investment c a n generate superior returns m a y be sufficient to leave s o m e doubt
about the validity of the efficient markets view but are
not strong e n o u g h to m a k e a persuasive case that socially responsible investors will in fact earn superior
returns. T h e obvious next step is to look at the empirical evidence concerning h o w existing socially responsible investment f u n d s perform relative to the market.
A simplistic approach to analyzing the evidence yields
mixed results. S o m e socially responsible investment
mutual f u n d s over some periods of time produce rates
of return in excess of other mutual f u n d s and the market. 1 0 However, other comparisons of socially responsible investment f u n d s find that they produce inferior
returns over s o m e periods of time. 11
Thus, finding an answer to the question of whether
socially responsible investing outperforms conventional
investing is not an easy task. T w o problems in particular
raise doubts about simplistic comparisons of socially responsible investment mutual fund results. First, stock
returns contain an element that is not predictable: that
is, random luck will influence the results. For example,
unexpected changes in consumer tastes may turn out to
favor industries with historically good (or bad) employee relations. T h u s , a socially responsible i n v e s t m e n t
f u n d that focuses on f i r m s with good e m p l o y e e relations may earn higher (or lower) returns for reasons totally unrelated to the f u n d ' s investment criteria.
Reserve Bank of Atlanta
DigitizedFederal
for FRASER
A second p r o b l e m in m a k i n g c o m p a r i s o n s is that
r i s k i e r s t o c k s s h o u l d p e r f o r m d i f f e r e n t l y than less
risky stocks. T h e p e r f o r m a n c e of riskier stocks is likely to be better during s o m e periods and worse during
o t h e r s r e l a t i v e to less r i s k y s t o c k s . F u r t h e r , r i s k i e r
stocks should provide higher returns over the long run
to c o m p e n s a t e investors for the greater risk. 1 2 T h u s ,
any analysis of the p e r f o r m a n c e of socially responsible investment f u n d s should take account of risk.
Sally Hamilton, Hoje Jo,
alyzed the performance of
ment mutual funds relative
a methodology designed to
and Meir Statman (1993) ansocially responsible investto other mutual funds using
take account of random luck
For socially responsible investment to make
a real difference in the world changes must
begin with firms' investment policies.
and differences in the riskiness of various portfolios.
T h e measure they use, Jensen's alpha, is based on the
capital asset pricing m o d e l ( C A P M ) . 1 3 T h e y f o u n d that
socially responsible investment mutual f u n d s started
prior to 1986 outperform comparable conventional mutual f u n d s and that socially r e s p o n s i b l e i n v e s t m e n t
m u t u a l f u n d s started in 1986 or later u n d e r p e r f o r m
conventional mutual f u n d s . H o w e v e r , in neither case
are the results statistically s i g n i f i c a n t , w h i c h m e a n s
that the differences in performance could be due to rand o m luck. T h u s , their results suggest that socially res p o n s i b l e i n v e s t m e n t m u t u a l f u n d s ' p e r f o r m a n c e is
neither better nor worse than funds that do not follow
socially responsible investment principles. T h e Hamilton, Jo, and Statman results are consistent with the prediction of the efficient markets hypothesis, which
suggests that n o p o r t f o l i o selection criteria, w h e t h e r
based on social values or a particular economic theory,
will produce consistently superior investment results.
Potential socially responsible investors m a y be disa p p o i n t e d that existing e v i d e n c e d o e s not show that
such a strategy p r o d u c e s superior returns. T h e y m a y
Economic
Review
3
take c o m f o r t , however, f r o m the evidence that neither
does it produce inferior returns. T h e next question facing potential socially responsible investors is whether
they can help improve the world through their investm e n t s while obtaining portfolio returns comparable to
conventional f u n d s on a risk-adjusted basis.
Effective Socially
Responsible Investment
For socially responsible investment to m a k e a real
difference in the world changes must begin with firms'
investment policies. Socially desirable firms must be enc o u r a g e d to e x p a n d by i n v e s t i n g m o r e in plant and
equipment than they would without socially responsible
investment, and not investing in socially undesirable
firms, at least as long as they follow undesirable policies, should force them to reduce their level of investment. In order to understand h o w purchases and sales of
stock motivated by socially responsible investment can
influence investment policy two issues need to be considered: (1) H o w do firms decide whether to invest in a
project? A n d (2) how could socially responsible investment change the results of the investment analysis?
Firms that maximize shareholder wealth should seek
to invest in projects whose returns equal or exceed the
rate of return r e q u i r e d by investors. 1 4 T h e p r e f e r r e d
method used to determine whether the project returns
are sufficiently high is called net present value (NPV). 1 5
T h e f o r m u l a for calculating net present value and an
e x a m p l e are presented in Box 1. Essentially, net present value works by comparing the initial investment in
a project with the adjusted value of future cash flows.
T h e adjustment procedure uses a discount factor to reduce the value of future inflows to compensate f o r the
time value of money and the risk of the project. A large
reduction in the value of future cash i n f l o w s — a high
discount f a c t o r — d i s c o u r a g e s firms f r o m investing in
new projects. 1 6 Conversely, a low discount factor will
encourage investment.
T h i s discussion suggests that socially responsible
investment might change f i r m s ' investment practices if
it could c h a n g e the discount factor applied to investm e n t decisions. Can socially responsible i n v e s t m e n t
change the discount factor? T h e answer to this question
turns out to be yes: if socially responsible investment
can increase a f i r m ' s stock price, it m a y decrease the
discount factor. C h a n g e s in stock price influence the
discount factor because the discount factor should dep e n d on a f i r m ' s cost of f u n d s . C h a n g e s in a f i r m ' s
http://fraser.stlouisfed.org/
Econom ic Review
4
Federal Reserve Bank of St. Louis
stock price influence the cost of equity to the firm and,
hence, its estimated cost of funds for future periods (as
m e a s u r e d by the w e i g h t e d a v e r a g e c o s t of c a p i t a l
[ W A C C ] ) . A higher stock price implies a lower discount f a c t o r applied to f u t u r e cash f l o w s and h e n c e
greater investment by the firm. Following the socially
responsible investment prescription of investing in socially desirable firms could therefore induce these
firms to increase their investment in new and expanded
projects. O n the other h a n d , r e d u c i n g a f i r m ' s stock
price may drive up the discount factor. T h e f i r m ' s higher W A C C would discourage further investment.
O n e caveat to this analysis should be noted. In order
for socially responsible investment to cause a c h a n g e
in a f i r m ' s investment policy the c h a n g e in the stock
price must be large and must persist over an extended
period of time. N P V calculations are subject to substantial measurement error in at least three areas: (1) the estimation of the expected cash f l o w s , (2) the estimation
of the cost of equity, and (3) the process of adjusting
for differences between the risk of the project and the
f i r m ' s a v e r a g e risk. F i r m s generally r e c o g n i z e these
m e a s u r e m e n t errors and try to o f f s e t t h e m by using
rules that are s o m e w h a t m o r e conservative than is implied by a straightforward calculation of N P V using
W A C C . For e x a m p l e , f i r m s m a y not calculate a new
W A C C daily to r e f l e c t c h a n g e s in their stock price.
Thus, small, temporary changes in share prices are unlikely to cause any change in a f i r m ' s investment policy.
Socially Responsible Investment and
C o m m o n Stock Prices
The above analysis suggests that socially responsible investment may be effective in changing f i r m s ' investment policies if it can cause a substantial, long-run
c h a n g e in their stock prices. T h u s , the q u e s t i o n b e c o m e s that of whether socially responsible investment
can c h a n g e a f i r m ' s stock price.
A n efficient markets perspective says that a f i r m ' s
stock price reflects all available information about a
s t o c k and that all s t o c k s earn only a rate of return
c o m m e n s u r a t e with their market risk. T h u s , under efficient markets, socially responsible investment must be
ineffective. Yet, as a n y o n e learns in one semester of
e c o n o m i c s , m a r k e t prices are set by supply and dem a n d . Since socially responsible investment policies
c h a n g e either the supply or d e m a n d f o r a stock, the
law of s u p p l y and d e m a n d s u g g e s t s that t h e p r i c e
should also c h a n g e . R e s o l v i n g this s e e m i n g c o n f l i c t
January/February 1995
Box 1
Project Valuation
c o u n t f a c t o r applied to the cash f l o w s . F o r e x a m p l e , sup-
Net Present Value
F i n n s that m a x i m i z e s h a r e h o l d e r wealth should s e e k
p o s e t h a t t h e d i s c o u n t f a c t o r is l o w e r e d t o 8 p e r c e n t .
to invest in projects w h o s e returns, a d j u s t e d f o r the t i m e
T a b l e 2 s h o w s that in this case p r o j e c t A retains a posi-
value of m o n e y and risk, equal o r e x c e e d the rate of re-
tive NPV,
turn r e q u i r e d by i n v e s t o r s . T h e m e t h o d u s e d t o a d j u s t
negative. T h u s , l o w e r i n g the d i s c o u n t f a c t o r has resulted
p r o j e c t s ' e a r n i n g s for the t i m e v a l u e of m o n e y and risk is
in a d d i t i o n a l i n v e s t m e n t a n d e x p a n d e d the s i z e of the
called net p r e s e n t value
firm. Conversely, s u p p o s e that Peachtree M u l t i m e d i a
(NPV).
7
NPV
w h e r e NPV
project B gains a positive NPV,
but C r e m a i n s
S o f t w a r e w a s considered to be socially irresponsible and
CFt
socially responsible investors w o u l d p r e f e r that it m a k e
5d+/•)'
n o n e w i n v e s t m e n t . In this c a s e an i n c r e a s e in the dis-
= net p r e s e n t v a l u e , T = u s e f u l life of the
p r o j e c t , t = n u m b e r of d i s c o u n t i n g periods, CF f = cash
c o u n t f a c t o r to 12 percent w o u l d m a k e all of the projects
have a negative
NPV.
f l o w at t h e e n d of period t, a n d r = cost of capital (see
Weighted Average Cost of Capital
Peterson 1994, 3 9 9 - 4 0 5 ) .
T h e m e t h o d of calculating NPV m a y b e illustrated by
T h e discussion of net present v a l u e s u g g e s t s that de-
a p p l y i n g it t o t h e t h r e e h y p o t h e t i c a l p r o j e c t s g i v e n in
termining the a p p r o p r i a t e d i s c o u n t f a c t o r is i m p o r t a n t in
T a b l e 1.' For c o n v e n i e n c e , a s s u m e that all three p r o j e c t s
project valuation. T h e appropriate discount factor de-
are of equal risk and that their risk is also equal to the av-
p e n d s o n the f i r m ' s cost of f u n d s . C a l c u l a t i n g the cost of
e r a g e risk of the f i r m P e a c h t r e e M u l t i m e d i a S o f t w a r e . In
f u n d s w o u l d be a trivial p r o c e s s f o r a f i r m that w a s all
all three c a s e s the e x p e c t e d cash f l o w s o v e r the life of the
d e b t - f i n a n c e d if there w e r e n o taxes. In this c a s e the cost
p r o j e c t s e x c e e d s the i n v e s t m e n t c o s t . H o w e v e r , b e f o r e
of f u n d s w o u l d b e the interest rate o n the debt. In a f i r m
i n v e s t i n g in a n y of t h e p r o j e c t s P e a c h t r e e M u l t i m e d i a
s u b j e c t t o t a x e s with a m i x t u r e of debt a n d e q u i t y the
Software should determine whether the expected cash
m e a s u r e m e n t b e c o m e s m o r e complicated. A firm that
f l o w s a d e q u a t e l y c o m p e n s a t e for the t i m e value of m o n -
seeks to m e a s u r e its cost of capital m u s t a d d r e s s three is-
ey and the riskiness of the projects. T h e net p r e s e n t value
sues: (1) the cost of its e q u i t y , (2) h o w it c a n a d j u s t f o r
calculation t a k e s a c c o u n t of the t i m e v a l u e and riskiness
the d i f f e r e n c e s in the tax t r e a t m e n t of debt a n d e q u i t y ,
by d i s c o u n t i n g f u t u r e cash f l o w s at the rate r. S u p p o s e ,
a n d (3) h o w it can a d j u s t f o r the p r o p o r t i o n of debt and
f o r e x a m p l e , that P e a c h t r e e M u l t i m e d i a S o f t w a r e d e c i d e s
equity c o n t a i n e d in its capital structure. 4
2
the appropriate discount f a c t o r is 10 percent. 3 In this c a s e
t h e NPV
o f p r o j e c t A is c a l c u l a t e d a s f o l l o w s ( w i t h
T h e r e are a variety of w a y s to calculate a f i r m ' s cost
of c o m m o n equity. 5 O n e widely used m o d e l is a d i v i d e n d
growth valuation model.6 The dividend growth model
r o u n d i n g to the nearest dollar):
says that the value of a stock should be the net present
NPV = - $ 1 0 , 0 0 0 + $ 4 , 0 0 0 / 1 . 1 0 + $4,500/(1.10) 2
v a l u e of its e x p e c t e d d i v i d e n d p a y m e n t s . F u r t h e r , the
v e r s i o n of t h e m o d e l u s e d m a k e s t h e s i m p l i f y i n g a s -
3
+ $4,100/(1.10) = $187.
s u m p t i o n that d i v i d e n d s are e x p e c t e d to g r o w at a c o n P r o j e c t A w o u l d be a c c e p t e d in this c a s e b e c a u s e it has a
stant rate forever. 7 For e x a m p l e , s u p p o s e that P e a c h t r e e
net present value of $ 1 8 7 , m e a n i n g that the value of its
M u l t i m e d i a S o f t w a r e is currently p a y i n g d i v i d e n d s of $7,
e x p e c t e d cash i n f l o w s a d j u s t e d f o r risk a n d the time val-
these d i v i d e n d s are e x p e c t e d to g r o w at a 2 percent a n n u -
ue of m o n e y e x c e e d s the required i n v e s t m e n t by $ 1 8 7 . In
al rate, a n d the f i r m ' s s t o c k price is $ 5 0 :
f i n a n c e terms, project A should be u n d e r t a k e n b e c a u s e it
is a positive NPV project. T h e net present values of proj-
r = Dr/P0
+ g = 7 / 5 0 + 2 % = 16%,
ects B and C are g i v e n in the s e c o n d and third c o l u m n s
of T a b l e 2. T h e a d j u s t e d value of the e x p e c t e d cash in-
w h e r e re = required return o n equity, Dt = dividends per
f l o w s f r o m t h e s e is less than t h e v a l u e of the r e q u i r e d
share of stock at time t, P0 = present value of a share of
outlay. T h e s e projects h a v e a n e g a t i v e NPV
stock = its price at time zero, and g = e x p e c t e d growth rate
and should
be rejected.
S u p p o s e that P e a c h t r e e M u l t i m e d i a S o f t w a r e is c o n -
of dividends per share. T h u s , the required rate of return on
Peachtree M u l t i m e d i a S o f t w a r e stock is 16 percent.
sidered to be socially responsible and socially responsible
Interest p a y m e n t s o n debt m a y be d e d u c t e d by firms
investors w o u l d like to see the f i r m g r o w m o r e rapidly.
w h e r e a s d i v i d e n d p a y m e n t s t o s t o c k h o l d e r s are not tax
Stock investors m a y not be able to c h a n g e the cash f l o w s
d e d u c t i b l e . 8 T h u s , t o d e t e r m i n e t h e c o s t of t h e t w o
to the p r o j e c t , but they m a y be a b l e to c h a n g e the dis-
sources o f f u n d s to the f i r m , an a d j u s t m e n t m u s t b e m a d e
Federal
Reserve Bank of Atlanta
continued on next page
Economic
Review
5
continued
from page 5
t o r e c o g n i z e t h e d i f f e r e n c e in tax t r e a t m e n t . T h e s t a n d a r d
a p p r o a c h is t o r e d u c e t h e interest r a t e t h e firm p a y s t o ref l e c t t h e f a c t that t h e g o v e r n m e n t s h a r e s in t h e c o s t o f
m i l k i n g t h e interest p a y m e n t (in t h e s a m e s e n s e that t h e
g o v e r n m e n t s h a r e s in t h e f i r m ' s r e v e n u e a n d its o t h e r
c o s t s of p r o d u c t i o n s u c h as w a g e p a y m e n t s t o w o r k e r s ) .
Suppose, for example, that Peachtree Multimedia Softw a r e c o u l d i s s u e n e w d e b t that y i e l d e d 8 p e r c e n t p e r y e a r
m i x of debt and equity, these costs need to be averaged
to d e t e r m i n e the f i r m ' s overall cost of capital. T h e
w e i g h t e d a v e r a g e c o s t o f c a p i t a l ( W A C C ) c a l c u l a t e s this
f i g u r e , as its n a m e s u g g e s t s , b y t a k i n g w e i g h t e d a v e r a g e s
of the cost of debt and equity where the weights equal
t h e p r o p o r t i o n of f u n d i n g t h a t t h e f i r m o b t a i n s f r o m e a c h
s o u r c e . If P e a c h t r e e M u l t i m e d i a S o f t w a r e is f u n d e d b y
6 0 p e r c e n t d e b t , t h e n its WACC
t o i n v e s t o r s a n d that P e a c h t r e e w a s in a 2 5 p e r c e n t tax
WACC
b r a c k e t . In this c a s e t h e a f t e r - t a x c o s t of d e b t t o P e a c h t r e e
= wd rd +wer=
m a y b e c a l c u l a t e d as
(60% • 6%)
+ (40% • 1 6 % ) = 10%,
is c a l c u l a t e d as f o l l o w s ( s e e P e t e r s o n 1994, 6 3 6 - 3 7 ) :
w h e r e wd = p r o p o r t i o n of the f i r m ' s f u n d i n g o b t a i n e d
r*d = rd( 1 - t ) = 8 % ( 1 - .25) = 6 % .
f r o m debt and w = proportion of the f i r m ' s f u n d i n g o b T h u s , t h e a f t e r - t a x c o s t of d e b t t o P e a c h t r e e M u l t i m e d i a
t a i n e d f r o m e q u i t y . 9 T h u s , P e a c h t r e e ' s WACC
is e q u a l t o
10 p e r c e n t . 1 0
S o f t w a r e is 6 p e r c e n t .
In g e n e r a l t h e a v e r a g e a f t e r - t a x c o s t s of d e b t a n d e q u i ty will d i f f e r . T h u s , if a f i r m ' s capital s t r u c t u r e c o n t a i n s a
Table 1
Project Cash Flows
Project A
Project B
Project C
$10,000
$10,000
$10,000
Y e a r 2
$4,000
$4,400
_
_
$3,700
$4,200
$3,600
$4,000
Year
$4,000
$4,050
$4,000
$2,400
$1,950
$1,600
Project A
Project B
Project C
initial discount factor = 1 0 %
$187
($122)
($416)
reduced factor for
socially responsible firm = 8 %
$559
$242
increased factor for
($163)
($466)
Initial investment outflow
Net cash inflows
Y e i r ,
3
Net cash flows
Table 2
Net Present Values
NPV given
($62)
($750)
socially irresponsible firm = 1 2 %
http://fraser.stlouisfed.org/
Econom
6
Federal Reserve Bank of St. Louis
ic
Review
J a n u a r y / F e b r u a r y 1995
Notes
1.The formula for calculating net present value is widely accepted, but there are important questions about its implementation. For example, how should cash flows from the
project be measured? And how should the value of the implicit option to defer the project be valued? Given the focus
of this article, the issue of how to measure cash flows may
be especially important in some cases. For example, failure
to incorporate future outlays by the firm to clean up environmental contamination will bias project selection in favor of
environmentally unsound projects. This research assumes
that these measurement issues have been correctly addressed
by individual firms, however, since the issue of how to measure both cash flows and the option of deferral for individual
projects are management issues that are generally outside
the control of investors.
2. These assumptions about project risk are not necessary for
the conclusions of this section. They are used only to simplify the analysis.
3. A discussion of how to calculate r will follow in the next
section. Note that in some cases a time varying discount rate
may be more appropriate. For example, r may equal 10 percent for the first period and 11 percent for the second period.
However, introducing this complication would not change
the conclusions of the analysis.
4.The calculation of the cost of capital may also be influenced
by changing the mix of debt and equity. As a part of the assumption that management maximizes shareholder wealth,
this article assumes that firms use the lowest cost mix of
debt and equity (referred to as optimal capital structure).
between efficient markets and the law of supply and
d e m a n d will show whether socially responsible investment m a y influence the stock price and, if so, when.
The key to reconciling the t w o views is to note that
supply and d e m a n d are, in e c o n o m i s t s ' terms, perfectly elastic in efficient markets. 1 7 That is, any imbalance
in supply or d e m a n d that threatens to cause the market
price to deviate f r o m the f i r m ' s underlying value will
produce an immediate change in the behavior of other
investors. For e x a m p l e , if an excess of d e m a n d by socially r e s p o n s i b l e i n v e s t o r s o c c u r s that t h r e a t e n s to
push the stock price a b o v e its correct value, other investors will reduce their d e m a n d for the stock or increase their supply to keep it f r o m exceeding its value
and b e c o m i n g undesirable for non-socially responsible
investors. T h e y would rather put their f u n d s in stocks
whose market price value is no m o r e than its intrinsic
e c o n o m i c value. T h e net result is that the stock price
will r e m a i n u n c h a n g e d . C o n v e r s e l y , an i n c r e a s e in
supply f r o m socially r e s p o n s i b l e investors d u m p i n g
socially undesirable firms will lead other investors to
increase their d e m a n d or decrease supply so that the
price is unchanged. 1 8
Federal
Reserve Bank of Atlanta
5. In order to simplify the analysis Peachtree Multimedia Software is assumed to have no outstanding preferred stock.
6. Another way of calculating the market's expected rate of return on a stock is by estimating the capital asset pricing
model (CAPM). In theory these methods should yield similar results. In practice the different methods may not yield
identical results, in part because of problems obtaining exact
values for some key parameters. This example is based on
the dividend growth model because the relationship between
stock price and required rate of return is easier to see in a
dividend growth model. Also, the CAPM may not be usable
for analyzing the impact of socially responsible investment
since that model assumes that stock prices are set in efficient
markets.
7.This assumption could easily be relaxed without changing
the conclusions.
8.This general statement is subject to some exceptions that are
not important for the purposes of this study.
9. Note that these weights are typically based on the book value of outstanding debt and equity, but market values could
be used instead. If book values are used, then retained earnings are considered part of owner's equity.
10.The formula for WACC comes from Peterson (1994, 653).
The one difference is that the term for preferred stock is
dropped from the equation because Peachtree Multimedia
Software is assumed to have no outstanding preferred stock
(w = 0 in Peterson's terminology).
F e w financial economists would contend that stock
prices are perfectly elastic. 19 Hence, the efficient markets
view is unlikely to hold exactly for all c o m m o n stock at
all times. However, the market for some stocks is huge, especially those of large funis such as the ones appearing in
the often-cited D o w J o n e s Industrial Average. T h e s e
stocks are followed by a large n u m b e r of analysts so that
investors have considerable information on these firms.
Moreover, trillions of dollars in international investment
funds are prepared to m o v e into or out of these stocks if
they deviate significantly f r o m the market's perception
of their value. Thus, socially responsible investors are
likely to have a difficult time influencing the investment
policies of very large firms because they are unlikely to
cause significant long-run deviations of their stock price
f r o m their economic value.
T h e best chances for socially responsible investors
to i n f l u e n c e f i r m s ' i n v e s t m e n t t h r o u g h t h e i r s h a r e
price m a y be by purchasing shares in smaller c o m p a nies that f o l l o w socially desirable policies. 2 0 Smaller
c o m p a n i e s are followed by f e w e r investors, and these
investors often h a v e f e w e r resources at their disposal. 21 Further, socially responsible investors are m o r e
Economic
Review
7
likely to be successful in bidding up the price of desirable stocks for extended periods of time than they are
in forcing d o w n the price of undesirable stocks over
long periods. 2 2 Investors may try to force a stock price
d o w n by selling their shares, but the effect on the stock
price m a y not be significant unless they o w n a large
block of shares. 2 3 Even if the stock price declines significantly, the drop is likely to be temporary as n o n socially responsible investors increase their purchases
of what they would perceive as an "undervalued" asset. 24
Although most fund managers do little short-selling,
an extremely aggressive socially responsible investor
or f u n d m a n a g e r could try to depress an undesirable
s t o c k ' s price via s h o r t - s e l l i n g . S h o r t - s e l l i n g o c c u r s
when investors who d o not own a stock borrow shares
from an existing owner to sell. A n investor w h o sells a
stock short m a y profit if the stock price subsequently
falls and the borrowed shares can be replaced at a lower price, but the short-seller stands to lose if the stock
price appreciates. Short-selling to exploit overvalued
stocks is generally more costly than buying a stock to
exploit undervalued shares for t w o reasons: (1) shortsellers must pay any dividends earned to the owner of
the stock they borrowed, and (2) short-sellers face restrictions on the use of the proceeds of their short-sale
to protect the owner of the borrowed shares. Socially
responsible investors m a y f i n d these costs excessive in
trying to use short-selling to permanently depress the
stock price of undesirable firms.
/triplications of Effective Socially
Responsible Investment for Stock Returns
A review of existing studies revealed t w o points of
view on the returns to socially responsible investors:
(1) the efficient markets view, which states that socially responsible investors should expect a m a r k e t rate of
return adjusted for risk, and (2) a pro-socially responsible i n v e s t m e n t view, which h o l d s that socially res p o n s i b l e f i r m s w i l l g e n e r a t e r e t u r n s s u p e r i o r to
irresponsible f i r m s over the long run. H o w e v e r , neither of these views explicitly takes account of the impact of effective socially responsible investment on a
f i r m ' s stock price and investments. If socially respons i b l e i n v e s t m e n t p o l i c i e s w e r e t o be e f f e c t i v e in
changing a f i r m ' s investment policies, there would be
clear implications for the returns earned by socially responsible investors.
A s Box 2 shows, in the short run, effective socially
responsible i n v e s t m e n t w o u l d be g o o d f o r investors
http://fraser.stlouisfed.org/
Econom ic Review
8
Federal Reserve Bank of St. Louis
w h o o w n e d the socially responsible stock b e f o r e investment boosted prices and w h o sold as soon as the
price increase driven by socially responsible investment stopped. The initial boost in the stock price enables investors to sell their stocks at a higher price and
thus earn a higher rate of return. However, in the long
run stock returns are lowered. T h e higher stock price
by itself implies an (eventually) l o w e r rate of return
when the f l o w of n e w socially responsible investment
m o n e y to the stock slows or stops; a reduction in new
p u r c h a s e s of socially r e s p o n s i b l e i n v e s t m e n t s t o c k s
will slow or stop the rate of appreciation in the f i r m ' s
stock price. Further, the percentage of return f r o m dividend p a y m e n t s will be reduced as the same dollar value is divided by a larger share price.
Conclusion
T h i s study e x a m i n e s three q u e s t i o n s about the financial implications of e f f e c t i v e socially responsible
investing in c o m m o n stocks—that is, socially responsible investment intended to c h a n g e f i l m s ' behavior.
T h e first question concerns what socially responsible
investors can d o to effectively influence f i r m s ' investment policies. T h e results suggest that to increase inv e s t m e n t by a f i r m socially r e s p o n s i b l e i n v e s t m e n t
must lower the f i r m ' s weighted average cost of capital
by significantly increasing the f i r m ' s stock price for a
period of time. C o n v e r s e l y , i n v e s t o r s s e e k i n g to red u c e a socially irresponsible f i r m ' s investment must
significantly reduce the f i r m ' s stock price for a time.
T h e s e c o n d question is, u n d e r what conditions, if
any, will the securities markets permit effective socially responsible investment? T h e analysis suggests that
although financial markets m a y not be perfectly efficient, socially responsible investors are likely to find it
d i f f i c u l t to c a u s e s i g n i f i c a n t , l o n g - r u n c h a n g e s in a
f i r m ' s stock price. T h e y are m o r e likely to influence
the stock price of smaller firms and to be m o r e effective in raising stock prices than in lowering them.
Third, what impact will socially responsible investment have on the p e r f o r m a n c e of portfolios that follow it? Effective socially responsible investment will
g e n e r a t e a b o v e - m a r k e t rates of return f o r i n v e s t o r s
w h o o w n e d targeted socially responsible stocks prior
to the b e g i n n i n g of socially responsible i n v e s t m e n t .
However, investors in socially responsible investment
stocks receive reduced returns once a stock b e c o m e s a
mature socially responsible investment, mature in the
sense that no new f u n d s are flowing into the stock. 2 5
January/February 1995
This analysis has two implications for fund man-
q u i r e m e n t s a l o n g the lines of those the
Community
agers and investors w h o want to change firms' behav-
R e i n v e s t m e n t A c t i m p o s e s on banks.26 T h i s article has
ior. F i r s t , t h e i n v e s t m e n t s t r a t e g y s h o u l d f o c u s o n
t w o implications for the question of i m p o s i n g socially
buying shares of small socially responsive firms. Sec-
responsible investment requirements on f u n d s hold-
o n d , to the extent that this strategy i n f l u e n c e s share
i n g s o f s t o c k . T h e f i r s t is that, u n l e s s p r o p e r l y
p r i c e s it m a y e a r n a b o v e - m a r k e t r a t e s o f r e t u r n in t h e
s i g n e d , s u c h r e q u i r e m e n t s m a y h a v e little i m p a c t o n
s h o r t r u n , b u t it is l i k e l y t o e a r n b e l o w - m a r k e t r a t e s in
f i r m s ' behavior. Further, to the extent that socially re-
de-
t h e l o n g r u n . O n e i s s u e n o t d i s c u s s e d a b o v e is t h e r i s k
s p o n s i b l e i n v e s t m e n t r e q u i r e m e n t s a r e e f f e c t i v e in i n -
i m p l i c a t i o n s of s u c h a f o c u s e d i n v e s t m e n t
fluencing firms, these requirements m a y reduce the
strategy.
J o h n H. L a n g b e i n a n d R i c h a r d A . P o s n e r ( 1 9 8 0 ) p o i n t
level of returns obtained by pension a n d m u t u a l funds.
out that socially responsible investment policies (effec-
T h u s , before any such requirements are i m p o s e d , both
tive or ineffective) m a y increase the riskiness of the
the social benefits of the requirements a n d the desir-
p o r t f o l i o b y l i m i t i n g its d i v e r s i f i c a t i o n . F u r t h e r , a f o c u s
ability of imposing a de facto tax on savings should be
on the stocks of small firms m a y also increase
considered.
risk.
S o m e analysts h a v e argued that mutual and pension
f u n d s s h o u l d b e s u b j e c t e d to s o m e sort of s o c i a l re-
Box 2
Implications of Effective Socially Responsible Investment for Investors
T h e i m p a c t of e f f e c t i v e s o c i a l l y r e s p o n s i b l e invest-
T h e increase in d e m a n d f r o m socially r e s p o n s i b l e in-
m e n t o n investor r e t u r n s m a y b e seen in an e x a m p l e in-
v e s t m e n t f u n d s that boosts W e C a r e ' s stock price to $ 5 5
volving a hypothetical f i r m , W e C a r e W i d g e t s . W e C a r e
d u r i n g the s e c o n d p e r i o d r e s u l t s in a r a t h e r substantial
W i d g e t s p r o d u c e s e x p e c t e d p r o f i t s of $ 6 per share, all of
holding period return of 103.3 p e r c e n t f o r the s e c o n d pe-
w h i c h are p a i d a s s t o c k h o l d e r d i v i d e n d s at the e n d of
riod, w h i c h is calculated as follows: HPR2
e a c h p e r i o d . T h e f i r m h a s n o debt, and it is a s s u m e d that
- 3 0 ) / 3 0 = 1 0 3 . 3 % . T h u s , a n y investors w h o p u r c h a s e d
= 6 / 3 0 + (55
its e a r n i n g s are equal to their e x p e c t e d value t h r o u g h o u t
the stock i m m e d i a t e l y b e f o r e t i m e 1 and w h o sold i m m e -
this e x a m p l e . It is also a s s u m e d that the required rate of
diately a f t e r t i m e 2 w o u l d h a v e m o r e than d o u b l e d their
return by non-socially r e s p o n s i b l e investors in the c o m -
i n v e s t m e n t in a single period.
p a n y will be constant t h r o u g h o u t the period. 1 T h e stock
T h e s l o w d o w n in n e w socially r e s p o n s i b l e i n v e s t m e n t
of W e C a r e W i d g e t s sells f o r $ 3 0 at both t i m e 0 and at
a f t e r t i m e 2 r e d u c e s the rate of return to investors. T h e
t i m e 1 (the e n d of the first period). H o w e v e r , i m m e d i a t e -
h o l d i n g p e r i o d return f o r the third period d r o p s t o 2 0 per-
ly a f t e r the end of the first period, the f i r m is r e c o g n i z e d
cent, in this c a s e exactly the s a m e rate of return e a r n e d
as being o u t s t a n d i n g in m e e t i n g its social responsibilities,
prior to socially r e s p o n s i b l e investors' d i s c o v e r y of W e
a n d socially r e s p o n s i b l e investors b e g i n p u r c h a s i n g the
C a r e W i d g e t s . M o r e o v e r , w h e n the f l o w of net n e w
stock. T h e stock price increases to $55 at t i m e 2, and it
f u n d s into the c o m p a n y ' s stock stops d u r i n g period 4, the
f u r t h e r increases t o $ 6 0 at t i m e 3. H o w e v e r , a f t e r t i m e 3
h o l d i n g p e r i o d rate of return to investors falls f u r t h e r to
the f l o w of a d d i t i o n a l s o c i a l l y r e s p o n s i b l e i n v e s t m e n t
10 percent: HPR4
f u n d s into the stock e n d s , and the stock price r e m a i n s at
ing that socially r e s p o n s i b l e investors stabilize the price
$ 6 0 at t i m e 4.
of W e C a r e W i d g e t s at $ 6 0 thereafter, f u t u r e h o l d i n g pe-
T h e m e a s u r e m e n t of return o n a stock should take a c -
= 6 / 6 0 + ( 6 0 - 6 0 ) / 6 0 = 10%. A s s u m -
riod returns will r e m a i n at 10 p e r c e n t .
c o u n t of both cash received by the investor (dividends or
interest) and any c h a n g e in the m a r k e t value of the asset:
Return
on a stock
P —P
D
= —
+ —>
p
Po
o
where P() = stock price at time 0, P, = stock price at time 1,
and Z} = dividends received at the end of period.-2 Thus, the
holding period return on W e Care's stock for the first period
is HPRI = 6/30 + (30 - 30)/30 = 20%. Given our assumptions, this 20 percent is also the WACC of W e Care Widgets.
e d e r a l Reserve B a n k of Atlanta
Digitized forFFRASER
Notes
1. All of the assumptions in this example are m a d e to simplify the discussion. Similar results could be s h o w n to
hold under more general assumptions.
2. See Peterson (1994, 260-65) for a discussion of measuring the return on a stock.
Economic
Review
9
Notes
1. Fierman (1994) argues that the management fees and trading costs incurred by many pension and mutual fund managers cannot be justified by their f u n d s ' performance. Her
analysis, which looks at money managers in general, finds
that only 26 percent of the 2,700 managers of equity funds
for pensions exceeded the S & P 5 0 0 over the last ten years.
2. See Kinder (1993) for a short overview of social investing.
3. The allocation of funds to social causes via socially responsible lending is only one of a number of ways that investment resources may be allocated in "socially s e n s i t i v e "
ways. See Srinivasan (1994) for a discussion of development lending where an information asymmetry exists between the borrower and the lender. Also see Tschinkel and
Wall (1994) for a discussion of investments that generate
public gains that can be (partially) captured by the government and shared with the private developer.
4. In a recent study Hamilton, Jo, and Statman (1993, 66) allude to this study's answers to the first and third questions.
H o w e v e r , their article d o e s not explain the logic behind
their arguments, presumably because underlying theory was
a s s u m e d to b e o b v i o u s to the f i n a n c e p r a c t i t i o n e r s f o r
whom the publication is intended. A recent working paper
by Knoll (1994) reaches the same conclusions to questions
1 and 3. His paper is somewhat more technical and provides
a more exhaustive discussion of the issues raised by socially
responsible investing.
5. For example, buyers of corporate debt yielding a belowmarket rate need to be assured that their investment will go
to expand the f i r m ' s operations rather than to expand the
shareholders' returns.
6. T h i s f o r m of the e f f i c i e n t m a r k e t s h y p o t h e s i s , which is
called the semistrong form of efficiency, is probably the
most widely applied. The two other forms of market efficiency are also interesting for s o m e applications. W e a k
f o r m efficiency merely requires that future stock returns
may not be predicted from prior returns. The strong form of
efficiency requires that stock prices reflect all available information, public and private. See Peterson (1994, 51-52).
7. Langbein and Posner (1980) discuss socially responsible investment in the context of efficient markets and the capital
asset pricing model (CAPM). The C A P M provides a specific mechanism for adjusting stock returns to take account of
risk. However, the efficient markets hypothesis does not depend on the C A P M or any other specific model of securities' returns.
8. Fama (1970, 1991) surveys the literature on market e f f i ciency.
9. The author is unaware of any empirical studies supporting
or refuting this claim.
10. Bruyn (1987, 13) provides a number of examples in which
investing based on socially responsible criteria would yield
superior results and in which socially responsible investment mutual funds outperformed the overall stock market
over selected time periods. Hylton (1992, 28-32) lists the
performance of a number of socially responsible investment
Econom ic Review
10
funds and notes that some of them have outperformed both
the S & P 500 index and the average of all mutual funds over
one- and five-year horizons ending in the early 1990s. She
also notes that some funds did not outperform the other indexes. See also Stoval (1992) and Kinder (1993) for some
evidence on socially responsible investment funds outperforming other mutual funds.
11. For example, Galen (1994) notes that over a three-month
period in 1994 a fund for " s i n n e r s " outperformed both a
fund for "saints" and the average of all mutual funds. The
sinners fund, Morgan Funshares, invested primarily in companies that specialize in industries such as alcohol, tobacco,
and gambling. See also Teper (1991) for statistics suggesting that using social criteria reduces a f u n d ' s performance.
12. See chapter 7 in Peterson (1994) for a discussion of risk and
return in finance.
13. The C A P M is a widely used model of stock returns. The
primar>' insight of the model is that investors should earn a
higher rate of return only for risks they cannot diversify
away ( n o n d i v e r s i f i a b l e risk). See Grinblatt and T i t m a n
(1994) for an analysis of alternative methods of analyzing
mutual fund performance.
14. An assumption maintained throughout the analysis is that
corporations' management and boards of'directors follow
policies designed to maximize shareholder wealth. If they
do not follow shareholder wealth maximization, then shareh o l d e r s m a y be able to use their v o t i n g p o w e r to f o r c e
changes in policies and thereby increase the f i r m ' s rate of
return. Shareholder initiatives to change firms' management
policies fall within the broad area of corporate governance,
a topic of considerable academic research interest in recent
years. Some of this research suggests substantial opportunities to improve many corporations' performance; however,
these studies have generally focused on issues unrelated to
that of following socially responsible investment principles.
This study will not address corporate governance issues for
two reasons: (1) little academic evidence exists to support
(or contradict) the hypothesis that firms following socially
responsible investment principals have superior earnings,
and (2) the issues raised by corporate governance questions
go far beyond the scope of this article.
15. An alternative is to calculate an internal rate of return (IRR)
for the project and c o m p a r e that with the f i r m ' s required
rate of return. N P V has several theoretical advantages over
IRR, but many businesses nevertheless use IRR. The choice
of N P V or IRR for analyzing projects is unimportant for the
purposes of this study. See chapter 9 of Peterson (1994) for
a discussion of the use of N P V and IRR for evaluating projects.
16. This discussion assumes that the investment project is normal in the sense that it requires a large initial investment
followed by a stream of cash inflows to the firm in future
years. The analysis would change if the project were to involve a sufficiently large cash outflow at the e n d — f o r example, a substantial expense at the end of the project to
January/February 1995
restore the environment to its original condition. In this case
a large reduction in the value of future cash flows (a large
discount rate) could actually encourage investment. Thus,
an increase in the real rate of interest (market interest rate
minus inflation) will have the effect of encouraging environmentally "dirty" projects that require clean-up in the future over projects that are environmentally "clean" from the
start.
17. Elasticity of demand (supply) refers to the change in quantity demanded (supplied) in response to a change in the market price.
18. This statement is based on the currently plausible assumption that most investors (or at least the marginal investors)
care only about the distribution of returns. If all investors
incorporated both return distribution and a similar set of social values in their pricing, then "economic values" would
depend at least in part on investors' social values. Note,
however, that merely having a majority of investors follow
socially responsible investment may not be sufficient for
economic values to incorporate social values. Non-socially
responsible investors may offset the socially responsible investors by holding less of the socially desirable stocks and
more of the undesirable stocks. Non-socially responsible investors would follow such a strategy if it promised a higher
risk-adjusted rate of return than holding stocks in proportion
to their market value (which in finance terms is referred to
as holding the market portfolio).
19. Indeed, a substantial body of academic finance literature has
arisen in recent years under the heading of " m a r k e t microstructure" to address the issue of how stock prices respond to changes in the order flow. See Kyle (1989) for an
example of a model in which prices are not set in perfectly
elastic m a r k e t s . F o r a general survey of the m a r k e t microstructure literature see O ' H a r a (1995) and the references
therein.
20. Bond investors may be able to lower a f i r m ' s W A C C by
buying primary issues of the firm at below-market yields.
21. Indeed, the very large investment funds often avoid investment in small capitalization stocks because the large funds
cannot buy or sell these stocks in sufficient quantity to influence their overall portfolio returns without causing large,
albeit often temporary, changes in the small f i r m ' s stock
price.
22. Although socially responsible investors may be able to bid
up the price of a stock in the short run, in order to maintain
the higher price socially responsible investors may need to
purchase virtually all of the stock that reaches the market in
the future. Investors w h o do not follow socially responsible
investing are likely to be reluctant to purchase a socially responsible stock after its price has increased because they
would be paying more for the stock than is indicated by its
fundamental value. (Equivalently, the expected rate of return on the stock may decline after its price has been bid up,
as is shown in the next section.)
23. The effectiveness of investors selling their shares may be
enhanced by coordinated efforts to have consumers boycott
the f i r m ' s products.
24. Scholes (1972) examines the impact of secondary stock offerings (large stock offerings that must be reported to the
Securities Exchange Commission) for New York Stock Exc h a n g e ( N Y S E ) stocks and f i n d s that these o f f e r i n g s reduced share price but the magnitude of the decline was very
small. More recently, Chan and Lakonishok (1993) examine
the impact of institutional trades on N Y S E and American
Stock Exchange ( A M E X ) stock returns. They find a larger
impact than Scholes and a bigger impact for buys than for
sells. H o w e v e r , they still find that these often very large
trades caused stock prices to move by less than 1 percent,
even for the smallest firms in their sample.
25. Munnell, Blais, and Keefe (1983) make essentially the same
point with regard to state investment in housing finance programs. Any state-sponsored program that e f f e c t i v e l y increases the supply of funds available to housing must result
in the state fund earning a below-market rate of return.
26. See Starobin (1993) for a discussion of the politics of imposing CRA-type requirements on nonbank intermediaries.
References
Bruyn, Severyn T. The Field of Social Investment. New York:
Cambridge University Press, 1987.
Chan, Louis K.C., and Josef Lakonishok. "Institutional Trades
and Intraday Stock Price Behavior." Journal of Financial
Economics 33 (April 1993): 173-99.
Fama, Eugene F. "Efficient Capital Markets: A Review of Theory and Empirical W o r k . " Journal of Finance 25 ( M a y
1970): 383-417.
. "Efficient Capital Markets: II." Journal of Finance 46
(December 1991): 1575-1617.
Fierman, Jaclyn. "The Coming Investor Revolt." Fortune 129,
October 31, 1994, 66-76.
Galen, Michele. "Sin Does a Number on Saintliness."
Week, December 26, 1994. 8.
Federal
Reserve Bank of Atlanta
Business
Grinblatt, M a r k , and Sheridan Titman. " A Study of Monthly
Mutual Fund Returns and P e r f o r m a n c e Evaluation Techniques." Journal of Financial and Quantitative Analysis 29
(September 1994): 419-44.
H a m i l t o n , Sally, H o j e Jo, and M e i r S t a t m a n . " D o i n g Well
While Doing Good? The Investment Performance of Socially Responsible Mutual Funds." Financial Analysts
Journal
(November/December 1993): 62-66.
Hylton, Maria O ' B r i e n . ' " S o c i a l l y Responsible' Investing: Doing G o o d v e r s u s D o i n g Well in an I n e f f i c i e n t M a r k e t . "
American University Law Review 42 (1992): 1-52.
Kaufman, George G., and Larry R. Mote. "Is Banking a Declining Industry? A Historical Perspective." Economic
Perspectives 18 (May/June 1994): 2-21.
Economic
Review
11
Kinder, Peter D. "Social Investing's Strength Lies in Readiness
to Deal with W o r l d ' s Tough Questions." Pension World 29
(April 1993): 10-12.
Knoll, Michael S. "Socially Responsible Investment and Modern Financial Markets." University of Southern California
Law Center Working Paper 94-13, December 1994.
Kyle, Albert S. "Informed Speculation with Imperfect Competition." Review of Economic Studies 56 (July 1989): 317-56.
Langbein, John H., and Richard A. Posner. "Social Investing
and the Law of Trusts." Michigan Law Review 79 (November 1980): 72-112.
Munnell, Alicia H., Lynn E. Blais, and Kristine M. Keefe. " T h e
Pitfalls of Social Investing: The Case of Public Pensions and
Housing." Federal Reserve Bank of Boston New
England
Economic Review (September/October 1983): 20-37.
O ' H a r a , Maureen. Market Microstructure
Theory. Cambridge,
Mass.: Basil Blackwell, Inc., 1995.
Peterson, Pamela P. Financial Management and Analysis. New
York: McGraw-Hill, Inc., 1994.
http://fraser.stlouisfed.org/
Econom ic Review
12
Federal Reserve Bank of St. Louis
Scholes, Myron S. " T h e Market for Securities: Substitution
versus Price Pressure and the Effects of Information on Share
Prices." Journal of Business 45 (April 1972): 179-211.
Srinivasan, Aruna. "Intervention in Credit Markets and Development L e n d i n g . " Federal Reserve Bank of Atlanta Economic Review 79 (May/June 1994): 13-27.
Starobin, Paul. " M a k e ' E m P a y . " National
1993, 1856-61.
Journal,
July 24,
Stoval, Robert H. " W h e n D o - G o o d e r s D o G o o d . " Financial
World 161, September 1, 1992, 68, 70.
Teper, Jeffrey A. " T h e Cost of Social Criteria." Pensions and
Investments, May 13, 1991, 34.
Tschinkel, Sheila L., and Larry D. Wall. " S o m e Lessons f r o m
Finance for State and Local Government Development Programs." Federal Reserve Bank of Atlanta Economic
Review
79 (January/February 1994): 1-10.
January/February 1995
i/nflation and
Inflation Forecasting:
An Introduction
Ellis W. Tal I man
nflation in the United States has b e e n moderate and relatively stable
f o r the past several y e a r s , h o v e r i n g at an a v e r a g e a n n u a l r a t e of
around 3 percent. F r o m the perspective of recent history, the currently modest rates of inflation are a vast i m p r o v e m e n t over the high and
o c c a s i o n a l l y d o u b l e - d i g i t i n f l a t i o n r a t e s of t h e 1970s a n d e a r l y
1980s. Since 1973 the inflation rate, m e a s u r e d as the rate of change f r o m
the year-ago quarterly C o n s u m e r Price Index (CPI), has averaged approximately 6 percent.
The author is an economist
in the macropolicy section of
the Atlanta Fed's
research
department. He wishes to
thank B. Frank King,
Eric Leeper, and Mary
Rosenhaum for
comments.
The author is solely
responsible for
contents.
Reserve Bank of Atlanta
Digitized forFederal
FRASER
Despite the relatively benign rates of inflation currently observed, inflation and the fear of inflation continue to penetrate the business world. Almost weekly, m a j o r m o v e m e n t s in the stock and bond markets are attributed
to e c o n o m i c data a n n o u n c e m e n t s that allegedly influence market perception
of the future inflation rate. Clearly, the lower rate of inflation observed recently does not greatly reduce the need or desire for inflation forecasts; the
costs of u n e x p e c t e d inflation on individual f i r m s and c o n s u m e r s r e m a i n
substantial e v e n at low levels of inflation.
Expectations about inflation are e m b e d d e d in planning decisions of all
kinds. Labor unions need inflation forecasts to help refine their w a g e dem a n d s , and firms planning for future expenses and expected revenues need
Economic
Review
13
to account for expected inflation. G o v e r n m e n t budgetm a k i n g has perhaps an even greater need for accurate
f o r e c a s t s of i n f l a t i o n , g i v e n t h a t S o c i a l S e c u r i t y
p a y m e n t s are linked directly with the i n f l a t i o n rate
m e a s u r e d by the CPT. Consider, for e x a m p l e , the implications of the following scenario. In 1993 the agg r e g a t e l e v e l of S o c i a l S e c u r i t y e x p e n d i t u r e s w a s
approximately $305 billion. S u p p o s e the g o v e r n m e n t
had budgeted Social Security expenditures for the next
ten years to increase at 3 percent per year but, because of
higher than expected inflation, the actual increase was
5 percent. T h e cumulative difference in the amount of
Social Security p a y m e n t s anticipated and the amount
paid in these ten years would be nearly $ 4 0 0 billion. 1
R e c e n t attempts at legislation to e n f o r c e a balanced
federal budget only heighten the importance of accurate inflation forecasts for planning expenditures and
revenues.
This article provides an overview of the effects of
inflation and the significance of inflation forecasting,
first considering s o m e of the reasons that businesses
m a y need to forecast inflation. T h e goal is to provide a
" c o n s u m e r s g u i d e " to forecast accuracy and evaluation. The discussion describes a selection of forecasti n g m e t h o d s , s o m e of w h i c h u s e b o t h s t a t i s t i c a l
models and forecaster j u d g m e n t , or "skill." The analysis e x a m i n e s as e x a m p l e s t w o statistical m o d e l s that
e m p l o y standard e c o n o m i c theory to suggest variables
that help predict inflation and reviews the subsequent
inflation forecasting models. Forecasts from each
m o d e l are then c o m p a r e d in a variety of w a y s with
simple univariate time series models that use only past
data on inflation to forecast the f u t u r e without reference to e c o n o m i c theory.
The Uses of Inflation Forecasts
T h e design of inflation forecasts reflects the varying needs and concerns of businesses, financial market
participants, government policymakers, and others
w h o are the end users of those predictions. Purely statistical m o d e l s (namely, time series m o d e l s ) e m p l o y
q u a n t i t a t i v e t e c h n i q u e s to extract i n f o r m a t i o n f r o m
observations of past data in order to generate a forecast w i t h o u t the u s e of e c o n o m i c theory. E c o n o m i c
m o d e l s use e c o n o m i c theory to j u s t i f y the choice of
e x p l a n a t o r y variables and help explain the b e h a v i o r
a m o n g the variables that underlie the forecast. In designing a formal forecasting model, the forecaster
c o n s i d e r s the u s e r ' s n e e d s in m a k i n g c h o i c e s a b o u t
Econom ic Review
14
several issues: (1) t h e t i m e h o r i z o n o v e r w h i c h the
model must forecast (one month, one quarter, one
year, and so forth), (2) the data frequency for the forecast (for example, monthly, quarterly, or annual data),
(3) the type of model (a time series model or a model
that e m p l o y s variables that hypothetically cause inflation), (4) the m e t h o d for evaluating forecast "accurac y " (for example, the root m e a n squared error of the
f o r e c a s t ) , the f o r e c a s t bias ( w h e t h e r on a v e r a g e the
forecasts over- or underpredict inflation), or the probability of predicting a turning point (say, f r o m low to
high inflation). 2
Models are specifically designed to fill the needs of
particular forecast users. For e x a m p l e , a private business like a m a n u f a c t u r i n g firm requires a significantly
different kind of inflation forecast than a p o l i c y m a k e r
or a financial market participant. Such businesses typically use f o r e c a s t s to help m o n i t o r input costs, reve n u e s f r o m output sales, the real (inflation-adjusted)
level of profits, and the like. Because businesses generally e m p l o y f o r e c a s t s with a m e d i u m to long-term
t i m e h o r i z o n — f r o m t w o to f i v e or ten y e a r s o u t —
quarterly or annual data are likely to be satisfactory
for their planning purposes. An additional consideration is that businesses rarely view inflation as an end
r e s u l t of t h e i r d e c i s i o n s or b e h a v i o r s so that t h e i r
n e e d s m a y be a d e q u a t e l y m e t by p u r e l y s t a t i s t i c a l
models. 3 The primary concern of a business inflation
f o r e c a s t will be f o r e c a s t a c c u r a c y , p e r h a p s as m e a sured by the typical evaluation criteria, but it may be a
combination of a n u m b e r of them. In general, m o d e l s
for businesses use the root m e a n squared error criterion because the criterion penalizes large errors and thus
helps avoid big misses in their forecasts. 4
A financial institution has quite different inflation
forecasting needs. For instance, it m a y want to predict
the probable future value of its fixed-rate loan portfolio. Similarly, financial portfolio m a n a g e r s use inflation forecasts to determine the best asset allocation for
m a x i m i z i n g real asset returns, perhaps substituting out
of assets w h o s e real returns erode amid inflation if the
forecast signals increasing inflation. T h e desired forecast horizons m a y range f r o m the short (monthly) to
the very long term (thirty years) for these forecast cons u m e r s , and m o d e l s designed f o r t h e m are likely to
e m p l o y m o s t of the available data frequencies. Both
i n t e r p r e t a b l e m a c r o e c o n o m i c and p u r e l y statistical
m o d e l s are used.
Inflation forecasts are also important to bond traders, w h o trade in markets that react quickly to new inf o r m a t i o n . F o r a f o r e c a s t i n g m o d e l to be u s e f u l in
their attempts to m a k e profits, bond traders m a y need
January/February 1995
to k n o w only the direction that i n f l a t i o n will m o v e
(relative to market anticipations) in the next inflation
rate a n n o u n c e m e n t . For e x a m p l e , the n e w s services
survey market participants' forecasts prior to an inflation rate a n n o u n c e m e n t and then publish the survey
average as the market forecast. Traders have an interest in a model that can improve on the average forecast. For such a short forecasting horizon, the m o d e l s
will use m o n t h l y data. T h e fact that only a limited
n u m b e r of m a c r o e c o n o m i c data are published monthly
perhaps constrains inflation prediction models that use
other m a c r o e c o n o m i c variables.
Because accuracy of inflation forecasts is an important consideration for bond trading operations, a n u m ber of forecasting techniques are likely to be used in
m o d e l s designed to m e e t their needs. As in models for
private businesses, the root m e a n square error m a y be
the key m e a s u r e of the a c c u r a c y of the bond trader
forecast because it helps reduce the likelihood of big
errors. J u d g m e n t is a n o t h e r key c o m p o n e n t of their
forecasts because of the absence of timely data. Any
descriptive short-term data relationships—that is, relationships that appear in the data but that may h a v e n o
deeper e c o n o m i c m e a n i n g — m a y help predict the next
m o n t h ' s inflation rate, and there is no need for any interpretability in the model.
Monetary and fiscal policymaking institutions, working without the profit motive of private businesses and
with social goals, h a v e a different objective in using
inflation forecasts. O n e motivation for a policymaker
is to attempt to minimize the social costs imposed by
inflation. E c o n o m i c m o d e l s chosen for analysis of inflation as well as for inflation f o r e c a s t i n g typically
take into account s o m e role for policy variables in the
inflation process. Policymakers typically assume that
policy choices affect inflation's behavior, and a m o d e l
of inflation that will be adequate for their purposes is
therefore likely to provide s o m e insight into policy's
potential impact on inflation. 5 The artificial m o d e l of
the inHation process then b e c o m e s the basis of the inflation forecasting model. The time horizon for an inflation forecast to serve policymakers should be long
e n o u g h to capture the effects of policy on the inflation
process, typically a s s u m e d to extend as f a r as t w o to
five years.
In most cases, the final inflation forecast n u m b e r s
for any of the entities mentioned above include adjustments to model forecasts that apply the j u d g m e n t of
the forecasters, that is, changes initiated to account for
information not contained in the e c o n o m i c or purely
statistical m o d e l . I n c l u d i n g such j u d g m e n t a l a d j u s t m e n t s in f o r e c a s t n u m b e r s m a k e s d i s t i n g u i s h i n g
Federal Reserve Bank of Atlanta
a m o n g f o r e c a s t i n g m o d e l s a m b i g u o u s b e c a u s e it is
difficult to correct for the f o r e c a s t e r ' s j u d g m e n t versus
the m o d e l implications. 6
.Macroeconomic Inflation
Forecasting Models
Of the two main types of inflation forecasting m o d e l s — p u r e l y statistical m o d e l s and e c o n o m i c ( m a c r o economic) models—economic models have some
characteristics that m a k e them typically more desirable.
As discussed, purely statistical models e m p l o y past inflation data to forecast future inflation, and there is no
additional explanation of the forecast. 7 Economic m o d els, on the other hand, are designed to be interpretable
so that movements in the explanatory variables explain
the inflation forecast. Also, inflation forecasts that use
data in addition to inflation data contain numerous correlations that purely statistical models cannot. 8
This article focuses on the forecasting p e r f o r m a n c e
of t w o simple e c o n o m i c approaches in describing the
inflation p r o c e s s . U s i n g s i m p l e f o r m u l a t i o n s a l l o w s
e m p h a s i z i n g the intuitive appeal of each m o d e l and
highlighting the contrast between them. While adding
other variables to the specification m a y i m p r o v e the
m o d e l s ' forecasting performance, doing so would hinder the direct inferences about the contribution of the
intuition of the m o d e l to forecasting p e r f o r m a n c e .
T w o representative m a c r o e c o n o m i c paradigms typically underlie c o m m o n empirical models of inflation and
appear comparably successful at forecasting inflation. 9
These two standard models incorporate interactions
a m o n g variables that are hypothesized to influence inflation significantly. 1 0 T h i s discussion e m p h a s i z e s the
m a i n intuitions f r o m the t h e o r e t i c a l m o d e l s as they
translate into the m o d e l specifications typically estimated for forecasting i n f l a t i o n . " Simple m o d e l s of inflation are e x a m i n e d with a f o c u s on w h a t the key
e l e m e n t s of e a c h a p p r o a c h c o n t r i b u t e to f o r e c a s t i n g
inflation.
O n e popular m a c r o e c o n o m i c m o d e l used for forecasting inflation is the so-called Phillips curve. 1 2 T h e
original Phillips c u r v e relation observed the negative
correlation of the rate of change in money wages in the
United Kingdom with the country's unemployment rate.
Over time, this empirical association has been similarly
applied to aggregate inflation and aggregate output.
T h e concepts of "potential output" (or potential gross
domestic product) and the "natural rate of unemploym e n t " help describe the aggregate application of the
Economic
Review
15
Phillips curve. T h e potential output measure indicates
the hypothesized a m o u n t of output that the e c o n o m y
could generate if it produced at full capacity; it is typically referred to as the full-employment level of output.
A n alternative description e m p l o y s what is called the
natural rate of unemployment—that is, the rate of u n e m ployment that persists in periods of healthy e c o n o m i c
g r o w t h , s u c h as u n e m p l o y m e n t r e s u l t i n g f r o m p o o r
w o r k e r - j o b m a t c h e s that result in resignations or t e m porary frictional u n e m p l o y m e n t and f r o m i n a d e q u a t e
w o r k e r t r a i n i n g r e l a t i v e to e x i s t i n g j o b o p p o r t u n i ties, k n o w n as structural u n e m p l o y m e n t . T h e Phillips
c u r v e r e l a t i o n s h i p in an a g g r e g a t e f o r m u l a t i o n s u g gests that an u n e m p l o y m e n t rate l o w e r than the natural rate of u n e m p l o y m e n t o r a g r o w t h rate of real
G D P that surpasses that of potential G D P is associated with a higher inflation rate. T h e concepts of potential output and the natural rate of u n e m p l o y m e n t are
nearly interchangeable in the Phillips curve inflation
model.
M o r e recent formulations of the Phillips curve app r o a c h s u g g e s t that the u n d e r l y i n g potential rate of
real growth represents the highest rate at w h i c h the
e c o n o m y could g r o w without increasing inflation. In
this view, output growth at this potential rate does not
exert inflationary pressure. If the e c o n o m y g r o w s at a
faster rate, firms face production capacity constraints
that force t h e m to hire additional labor and to w o r k
capital more intensively. In this process, firms bid up
the price paid to production factors (namely, wages),
and production costs increase. Firms pass the supply
price increases through to the final goods prices consumers face, causing inflationary pressures to increase.
T h e basic idea behind the Phillips curve is that if demand causes output to exceed the measure of potential
output, the economy gets stretched beyond its capacity
for noninflationary growth. T h e prices of then-scarce
p r o d u c t i o n f a c t o r s are bid u p , a n d the r e s u l t is increased prices that lead to reduced output d e m a n d . In
contrast, if output growth is below potential, inflationary pressures are reduced, and inflation should decline.
T h e d i f f e r e n c e b e t w e e n the real e c o n o m y ' s potential
output and actual output is referred to as the " g a p " or,
more specifically, the G D P gap.
Empirical m o d e l s that apply the Phillips curve intuition typically emphasize the central role of the G D P
gap in the inflationary process. 1 3 T h e empirical tests in
the research reported on here allow in the prediction
equation only lagged values of all variables that in the
initial s p e c i f i c a t i o n are c o n t e m p o r a n e o u s l y d e t e r m i n e d along with inflation. T h e n u m b e r of inflation
lags is limited to four. 1 4 T h e estimated relationship is
Econom ic Review
16
AP/ = a 0 + B ,(>',_, - y p M ) + 6 2 ( A 7 M - A y ^ )
The Phillips curve inflation forecasting m o d e l s often e s t i m a t e s i n g l e - e q u a t i o n m o d e l s like the o n e
above, suggesting that these simple m o d e l s capture the
key relationships in the data that determine inflation.
But single-equation m o d e l s are inherently partial equilibrium models, which hold m a n y other variables constant by a s s u m p t i o n , failing to account explicitly for
monetary or fiscal policies and their potential effects
on inflation. Also, the m o d e l is essentially a d e m a n d based model; potential output reflects the level of output that the e c o n o m y can supply without inflation, and
d e m a n d c o n d i t i o n s d e t e r m i n e w h e t h e r i n f l a t i o n increases or decreases.
T h e second m a c r o e c o n o m i c m o d e l presents an alternative view of the inflation process. The traditional
monetarist approach relies on the intuition that observations of past growth rate in the m o n e y supply predict the l o n g - r u n i n f l a t i o n rate. M o n e t a r y i n f l a t i o n
models suggest the following simple relationship:
Ap=f(MrAMl_]...Ml_,X
w h e r e Ap t is the inflation rate for period t and AM/ is
the m o n e y growth rate for period t. T h e simple relationship suggests that growth in the aggregate supply
of n o m i n a l money determines the inflation rate. Inflation is t h e r e f o r e s o m e f u n c t i o n of c u r r e n t and past
m o n e y growth measures.
M o s t empirical applications of the monetarist intuition s u g g e s t that the r e l a t i o n s h i p b e t w e e n m o n e y
g r o w t h a n d s u b s e q u e n t i n f l a t i o n is a s i n g l e l i n e a r
equation: the past and current growth rates of m o n e y
translate directly into subsequent inflation. 1 '' Empirical
estimations (and forecasts) of monetarist forecasting
m o d e l s depend largely on the choice of monetary aggregate used to measure m o n e y growth, typically M l ,
M 2 , or the monetary base. T h e appropriate monetary
measure (monetary aggregate) as well as the n u m b e r
of lagged observations of m o n e y growth for the inflation equation (what constitutes the " l o n g run") are not
suggested by theory but significantly affect the forecast p e r f o r m a n c e of the model.
T h e r e s u l t s in Y a s h P. M e h r a ( 1 9 8 8 ) , D a v i d J.
Stockton and J a m e s E. G l a s s m a n (1987), and William
Reichenstein and J. Walter Elliot (1987) suggest that
this simple monetarist formulation forecasts relatively
poorly regardless of the choice of monetary aggregate.
R a t h e r than u s e s u c h a f o r m u l a t i o n , the present research e m p l o y s an inflation forecasting model inspired
January/February 1995
by a simple m o n e y d e m a n d relationship presented in
E u g e n e F. F a m a (1982) and in a more extensive f o r m
in Elliot and Reichenstein (1987). T h e m o d e l i n g strategy bears a direct resemblance to the standard monetarist model because of the appearance of a monetary
aggregate in the specification. 1 6 However, the f r a m e work includes several other variables that help explain
the inflation process. For this article, the d e m a n d for
real balances (the m o n e y supply deflated by the price
level) is
Md/P
= f ( y , i).
The d e m a n d for real m o n e y balances increases as the
level of real activity increases, that is, as output grows.
Conversely, increases in the rate of interest raise the
opportunity cost of holding m o n e y balances (assumed
to be non-interest-bearing) instead of interest-bearing
assets and thus l o w e r the d e m a n d f o r real balances.
T h i s s i m p l e relationship p r o v i d e s the u n d e r p i n n i n g s
f o r another inflation forecasting model that uses a single linear equation, the m o n e y d e m a n d inflation forecasting equation.
The intuition underlying the m o n e y d e m a n d inflation forecasting equation is that a rate of m o n e y supply g r o w t h that is t h e s a m e as t h e r a t e of g r o w t h
d e t e r m i n e d by i n t e r e s t rate m o v e m e n t s and o u t p u t
( m o n e y d e m a n d ) will not increase inflation. W h e n the
m o n e t a r y a g g r e g a t e g r o w t h rate e x c e e d s the rate at
which m o n e y d e m a n d g r o w s (determined by the interest rate and o u t p u t m o v e m e n t s ) , the e x c e s s m o n e y
growth a f f e c t s the price level variable and results in
inflation. 1 7 T h e m o n e y d e m a n d model describes a simple m e c h a n i s m of inflation generation but does not attempt to uncover variable interactions that underlie the
general intuitive story.
Empirical specifications of the model vary, depending on the forecasting problem at hand. For the purposes of this study, the estimated m o n e y d e m a n d model
has four lags of the explanatory variables, m e a n i n g that
the f o u r m o s t recent v a l u e s of the e x p l a n a t o r y variables are included in the estimated equation. T h e initial specification of the m o n e y d e m a n d inflation model
has n o lags of the dependent variable. By adding four
lags of inflation to the right-hand side of the equation,
t h e m o d e l b e c o m e s m o r e c o m p a r a b l e to b o t h t h e
Phillips curve m o d e l and the time series model specification (see below), each having in c o m m o n four lags
of inflation as explanatory variables.
E s t i m a t e s of the s i n g l e - e q u a t i o n m o n e y d e m a n d
m o d e l of i n f l a t i o n s u f f e r f r o m c r i t i c i s m s s i m i l a r to
those aimed at the simple Phillips curve model, name-
Federal
Reserve Bank of Atlanta
ly, that the model is partial equilibrium and cannot isolate the underlying sources of shocks that generate inflation in the economy. T h e f r a m e w o r k is exclusively
for a m o n e y d e m a n d model and cannot separately acc o u n t f o r a g g r e g a t e supply s h o c k s like the oil price
s h o c k s o b s e r v e d in the 1 9 7 0 s . A n d e v e n t h o u g h a
monetary aggregate m e a s u r e is used in the model, the
data m e a s u r e d o e s not c a p t u r e c h a n g e s in m o n e t a r y
policies, often referred to as " s h o c k s . " Neither can the
simple m o d e l isolate the source of fiscal policy shocks
that affect the m o v e m e n t s in the explanatory variables.
This inability of partial equilibrium m o d e l s to isolate
shocks severely limits the degree to which the m o d e l s
m a y be interpretable—that is, they are not structural.
T h e s e s i n g l e - e q u a t i o n m o d e l s a s s u m e that the e x planatory variable values are s o m e h o w known before
inflation is forecast.
T h e m a c r o e c o n o m i c model forecasts will be c o m pared with those of purely statistical models. Univariate statistical m o d e l s , needing no other variables than
inflation in order to produce an inflation forecast, are
easy to use. Results c o m p a r e d with e c o n o m i c m o d e l
forecasts indicate w h e t h e r either of the theory-based
f o r e c a s t i n g t e c h n i q u e s i m p r o v e s i g n i f i c a n t l y on the
statistical m o d e l forecasts, which rely on past inflation
data only. For simplicity, a simple autoregressive m o d el of inflation with f o u r lags of inflation is estimated: 1 8
Ap! = a0 + l4j=lKjApi_j
+ el
M o d e l Estimation and
Forecast Evaluation
All the single-equation m o d e l s are estimated over
the data sample f r o m the fourth quarter of 1960 to the
third quarter of 1994. 1 9 Certain s h o r t c o m i n g s of the
m o d e l s can be detected f r o m the full s a m p l e estimations. Specifically, the m o n e y d e m a n d s p e c i f i c a t i o n
without lagged values of inflation suffers f r o m autocorrelated errors, that is, errors that are correlated with
past errors, indicating a problem of persistent prediction errors. 2 0 S u c h a pattern in forecast errors suggests
a serious problem in the forecast m o d e l design.
S o m e forecast evaluation p r o c e d u r e s e m p l o y e d in
the e c o n o m i c s l i t e r a t u r e m a k e s t r o n g , artificial a s s u m p t i o n s about the availability of data on variables
other than inflation. For e x a m p l e , in what is described
as dynamic, out-of-sample forecasting, single-equation
m a c r o e c o n o m i c f o r e c a s t i n g m o d e l s m a y e m p l o y the
actual values of the explanatory variables to generate
Economic
Review
17
the forecasts. In forecasts of one period into the future,
this strategy is adequate if the data on all explanatory
variables are available before the inflation measure. In
real time—that is, forecasting today what will not be
k n o w n until the f u t u r e — u s i n g future values that could
not be k n o w n at the time of the forecast is uninformative a b o u t the m o d e l ' s f o r e c a s t a c c u r a c y . T h e u n known values of the explanatory variables should be
forecast as well to generate a m o r e realistic test of a
m o d e l ' s forecasting accuracy.
The first forecasting exercise for the inflation m o d els is to predict the inflation rate that will take place
one quarter into the future. To investigate the relative
accuracy of the m o d e l s , o u t - o f - s a m p l e forecasting is
used in which the model is estimated over the sample
period up until the first forecasting period, in this case,
the fourth quarter of 1972. 21 T h e n , the model is forecast o n e p e r i o d into the f u t u r e , t h e first q u a r t e r of
1973. That forecast value is stored for later comparison with actual data. T h e estimation s a m p l e is then
updated to include the first quarter of 1973, and the
forecast process is repeated. A f t e r the model produces
Table 1
Forecast Evaluation One Quarter into the Future
Mean Error
Root Mean
Squared Error
Theil U
Forecast Sample: 1 9 7 3 : 1 - 1 9 8 3 : 4
Phillips Curve
Money Demand Plus
Time Series
Money Demand
.426
-.147
.179
.874
2.263
2.135
2.413
2.478
.889
.839
.948
.974
Forecast Sample: 1 9 8 4 : 1 - 1 9 9 4 : 3
Phillips Curve
Money Demand Plus
Time Series
Money Demand
.097
-.537
-.235
-1.940
1.603
1.788
1.595
2.850
.938
1.026
.933
1.666
Full Sample: 1973:1 - 1 9 9 4 : 3
Phillips Curve
Money Demand Plus
Time Series
Money Demand
.275
-.279
-.017
-.473
Econom ic Review
18
1.983
1.980
2.065
2.670
.907
.905
.944
1.220
forecasts for all periods in the forecast sample, the series of forecast values can be compared with actual inf l a t i o n data. B e c a u s e the e s t i m a t e d m o d e l s i n v o l v e
only lagged variables as predictor variables, the forecasts of each model can be c o m p a r e d with a real-time
forecasting problem in w h i c h inflation is forecast using only available data. 2 2 T h e results of the exercise
are in Table 1.
T h e evaluation criteria compare the forecast values
with the actual value of inflation and generate summary
m e a s u r e s of forecast accuracy. A s m e n t i o n e d above,
the root m e a n squared error is often the key m e a s u r e
of forecasting accuracy. However, one can also use the
m e a n e r r o r to g a u g e w h e t h e r the f o r e c a s t i n g m o d e l
generally over- or underpredicts the inflation rate. Depending on h o w the forecaster perceives the cost of errors, the m e a n absolute error m a y be preferred to the
root mean squared error. The mean absolute error does
not w e i g h a large e r r o r as h e a v i l y as d o e s the root
m e a n squared error (which imposes a quadratic cost of
error function).
Although all evaluation statistics are listed for the
f o r e c a s t s , the discussion will f o c u s on a m e a s u r e of
relative accuracy, the Theil U statistic. T h e Theil U
statistic represents the ratio of the root m e a n squared
error of the given forecasting model to the root m e a n
squared error of a naive " n o - c h a n g e in the dependent
v a r i a b l e " ( m e a n i n g , " t h e s a m e i n f l a t i o n rate as o b served last time") model, o n e that an individual naive
about e c o n o m i c theory can e m p l o y as a f o r e c a s t i n g
mechanism. As a criterion for a useful model, one
w o u l d h o p e that an inflation f o r e c a s t i n g m o d e l c a n
outperform this naive forecast. If the Theil U statistic
is greater than one, then the naive m o d e l o u t p e r f o r m s
the g i v e n f o r e c a s t i n g m o d e l . Values of the Theil U
statistic c l o s e to o n e s u g g e s t that the m o d e l u n d e r
scrutiny p e r f o r m s no better than a no-change forecast.
O n the other hand, values of the Theil U statistic less
than o n e i m p l y that the f o r e c a s t i n g m o d e l f o r e c a s t s
more accurately on average than the naive model. The
article f o c u s e s on the Theil U statistic as a w a y to
s u m m a r i z e and c o m p a r e root m e a n squared error, the
chosen criterion to distinguish a m o n g rival models. 2 3
Table 1 shows that the Phillips curve model and the
m o n e y d e m a n d plus inflation lags model perform similarly over the forecast subperiod f r o m the first quarter
of 1973 to the fourth quarter of 1983 as well as over
the entire f o r e c a s t s a m p l e . F o r the f u l l s a m p l e , the
Theil U statistics are similar and less than one, suggesting improvement over the simple "no-change"
forecast and slight improvement over the simple time
series m o d e l . A l s o , the P h i l l i p s c u r v e m o d e l h a s a
January/February 1995
m e a n error that is positive (underpredicting inflation
on average), whereas the m e a n error of the m o n e y dem a n d m o d e l is of s i m i l a r size but of o p p o s i t e sign
( o v e r p r e d i c t i n g ) . In c o n t r a s t , the m e a n e r r o r of the
time series m o d e l is close to zero for the full forecast
sample.
For the latter s a m p l e (the first quarter of 1954 to
the third quarter of 1994), the Phillips curve model has
a lower Theil U statistic than the m o n e y d e m a n d and a
smaller mean error that is close to one. O v e r this subsample, the time series m o d e l also appears to outperf o r m the m o n e y d e m a n d model, having a lower mean
e r r o r and a l o w e r T h e i l U statistic. In c o n t r a s t , the
Theil U statistic of 1.666 for the m o n e y demand m o d el without inflation lags suggests that the model forecasts significantly worse than a no-change forecast in
the latter forecast sample. 2 4 T h e m e a n error o f - 1 . 9 4 0
indicates that the m o n e y demand m o d e l (without inflation lags) is poorly specified, and it is not e x a m i n e d
further.
The next exercise for the m o d e l s is to forecast f o u r
quarters into the f u t u r e . T h e longer forecast horizon
r e q u i r e s the u s e of d a t a that in a r e a l - t i m e f o r e c a s t
would be unavailable. In this procedure the model is
estimated over the sample period, forecast four quarters into the future, and those forecast n u m b e r s are acc u m u l a t e d into o n e forecast of the a v e r a g e inflation
rate expected to persist over the next year. The process
is iterated as above to create a series of one-year ahead
forecast n u m b e r s . It is important to note, however, that
the n u m b e r of nonoverlapping data points or "observations" is m u c h smaller (the forecast sample divided by
four, or twenty-two yearly observations [if 1994 is included]) than in the one-step ahead forecast exercise. 2 5
There are t w o sets of results for the model evaluation statistics. The first set is evaluation statistics f r o m
a static forecast that a s s u m e s that the actual values of
data in the forecasting m o d e l are k n o w n (essentially,
g e t t i n g n e w i n f o r m a t i o n on t h e a c t u a l e x p l a n a t o r y
variables as the prediction extends further into the f u ture, but information that is unavailable f o r forecasting
in a real-time setting). T h e second set of results allows
partial d y n a m i c s — t h a t is, all the m o d e l s have lagged
inflation as an explanatory variable so that the model
forecasts of inflation are used as predictors, but all other explanatory variables e m p l o y actual data values in
the prediction equation. A fully dynamic model generates forecast values for future observations of the explanatory variables using the interrelationships a m o n g
the data to forecast them. T h i s approach contrasts with
partially d y n a m i c m o d e l s that use f u t u r e actual values,
which are obviously unavailable in a real-time forecast
e x e r c i s e . F o r e c a s t e v a l u a t i o n statistics are listed in
Table 2. 2 6
The most noticeable feature of the data in Table 2 is
how m u c h smaller the forecast error statistics are for
the static forecasts than f o r the partially d y n a m i c forecasts. T h e Theil U statistics for the static forecasts are
Table 2
Forecast Evaluation One Year into the Future
Mean Error
Static
Partially
Dynamic
Root Mean Squared Error
Static
Theil U
Partially
Dynamic
Static
1.943
1.923
2.256
.463
.480
.457
.855
.847
.993
1.292
1.649
1.472
.497
.685
.491
.940
1.199
1.071
Partially
Dynamic
Forecast Sample : 1 9 7 3 : 1 - 1 9 9 4 : 3
Phillips Curve
Money Demand Plus
Time Series
.277
-.335
-.063
.449
-.584
-.137
1.053
1.089
1.038
Forecast Sample: 1 9 8 4 : 1 - 1 9 9 4 : 3
Phillips Curve
Money Demand Plus
Time Series
Federal Reserve Bank of Atlanta
.049
-.447
-.254
.080
-.737
-.514
.683
.942
.676
Economic
Review
19
a l m o s t half t h o s e of the partially d y n a m i c f o r e c a s t
Theil U statistics, suggesting that the restriction of e m ploying a forecast of inflation rather than future actual
values of inflation almost doubles the variance of the
forecast. Another point worth noting is that the m e a n
errors are uniformly smaller in the static forecast exercise than in the partially dynamic forecasts, indicating
that the addition of some d y n a m i c s allows the m o d e l s
to stray m o r e often.
In the partially d y n a m i c results, both the Phillips
c u r v e and m o n e y d e m a n d w i t h i n f l a t i o n l a g s h a v e
Theil U statistics less than one, clearly better perform a n c e than the no-change forecast and s o m e w h a t better than the t i m e series m o d e l . F o r the f u l l s a m p l e
results, the mean errors of the respective models are of
opposite signs, 0.449 for the Phillips curve and - 0 . 5 8 4
f o r the m o n e y d e m a n d m o d e l s , s i m i l a r to the o n e quarter-out forecasts. Charts 1 and 2 present the forecast errors for the m o n e y d e m a n d and Phillips curve
m o d e l s , r e s p e c t i v e l y , o v e r the f u l l f o r e c a s t s a m p l e .
Positive forecast errors for both m o d e l s appear largest
around the t w o oil shocks ( 1 9 7 3 - 7 4 and 1979-80), ind i c a t i n g u n d e r p r e d i c t i o n of i n f l a t i o n and that t h e s e
simple demand models cannot adequately account for
supply shocks. Although both m o d e l s overpredict inflation during the 1975-76 recession and in the 1981-
Econom ic Review
20
83 r e c e s s i o n , the m o n e y d e m a n d m o d e l a p p e a r s to
have persistent overpredictions of inflation throughout
the e a r l y 1980s, a p e r i o d in w h i c h m a n y o b s e r v e r s
noted a d r a m a t i c shift in monetary policy. S o m e observers viewed the c h a n g e in Federal Reserve policy
(and possibly policy goals) in 1979 as a f f e c t i n g the
structure of the r e l a t i o n s h i p s a m o n g the data. 2 7 T h e
single-equation m o n e y d e m a n d model appears unable
to account for the c h a n g e in monetary policy and the
resulting effects on inflation.
In the shorter forecast sample f r o m 1984:1 to 1994:3,
the m o n e y d e m a n d specification has a Theil U statistic
of approximately 1.2, suggesting that the model forecasts less a c c u r a t e l y than the " n o - c h a n g e " f o r e c a s t .
Also, the m e a n error o f - 0 . 7 3 7 indicates that the money d e m a n d m o d e l overpredicts inflation on average,
which is undesirable if the forecast objective is an unbiased f o r e c a s t . O v e r this later s a m p l e , the Phillips
curve model appears to forecast m o r e accurately than
the alternative m o d e l s and the " n o - c h a n g e " forecast.
A s m e n t i o n e d above, there are unrealistic a s s u m p tions about data availability that are implicit in these
results. T h e use of actual data for the future observations of the e x p l a n a t o r y variables in-the f o r e c a s t i n g
equation offers an unfair advantage to the macroecon o m i c f o r e c a s t i n g m o d e l s relative to the t i m e series
Chart 1
Inflation Forecast Errors for Money Demand Models
(Static,
Partially
Dynamic,
and Dynamic
Forecasts)
January/February 1995
model. Given that advantage, the poor p e r f o r m a n c e of
the m o n e y d e m a n d model in the latter sample suggests
e v e n w e a k e r f o r e c a s t r e s u l t s w h e n the e x p l a n a t o r y
variables are forecast. T h e above results suggest that
the relationships underlying the m o n e y d e m a n d model
m a y h a v e weakened or changed in the latter sample,
perhaps as a result of the monetary policy shift, financial innovation, or both. 2 8
.Representative Dynamic Models
of Inflation
T h e s i m p l e l i n e a r f o r e c a s t i n g m o d e l s of b o t h
m a c r o e c o n o m i c approaches fail to specify h o w to generate values of the right-hand-side variables w h e n the
m o d e l forecasts out-of-sample, that is, when there are
no data available, as in a forecast two or three periods
into the future. A m o r e realistic procedure is to use an
explicitly d y n a m i c model to forecast inflation out-ofsample. Unfortunately, the simple models have no
suggestions about h o w to specify forecasting m o d e l s
for the explanatory variables in the m o d e l . Stockton
and Glassman (1987) e m p l o y e d time series specifications to forecast values of the explanatory variables in
their comparison of inflation forecasting models. For
this article, forecasts of the Phillips curve m o d e l were
generated by s p e c i f y i n g simple equations for the
growth in n o m i n a l G D P and the level of real G D P and
then i m p o s i n g identities to create the G D P g a p {ytX
- ypt_x) and the n o m i n a l g r o w t h g a p (AF f , - A_y/jM)
variables. 2 9 F o r the m o n e y d e m a n d m o d e l , a v e c t o r
a u t o r e g r e s s i o n (VAR) is e s t i m a t e d that i n c l u d e s the
m o n e y d e m a n d plus inflation lags as one equation of
the d y n a m i c system. T h e VAR model generates inflation forecasts in a d y n a m i c setting that m o r e closely
mimics the real-time forecasting p r o b l e m / "
In a v e c t o r a u t o r e g r e s s i o n , e v e r y variable on the
right-hand
side has a prediction e q u a t i o n a s s o c i a t e d
with it. T h u s , by design the model p e r f o r m s a truly dyn a m i c forecast of inflation o u t - o f - s a m p l e b e c a u s e it
f o r e c a s t s all v a r i a b l e s in the system. 3 1 VAR m o d e l s
f o r e c a s t in a p s e u d o - r e a l t i m e setting. T h a t is, they
forecast values of the explanatory variables implicitly
in a d y n a m i c forecast of inflation. T h e VAR generates
forecast values of the variables in the inflation equation, and the accuracy of the inflation forecasts relies
on h o w accurate the VAR forecasts the other variables
in the system. 3 2 T h e forecast evaluations of the VAR
present m o r e realistic forecast accuracy statistics f o r
the d y n a m i c forecasts of the average inflation rate f o r
Chart 2
Inflation Forecast Errors for Phillips Curve Models
(Static,
Partially Dynamic,
and Dynamic
Forecasts)
Percent
Federal
Reserve Bank of Atlanta
Economic
Review
21
next year. T h e m o n e y d e m a n d m o d e l with inflation
lags is essentially the equation that forecasts inflation
f r o m a vector autoregression m o d e l that contains all
the variables in the forecasting equation (that is, interest rates, output growth, monetary aggregate growth,
and inflation). These results appear in Table 3.
i n f l a t i o n c o m p o u n d s the initial f o r e c a s t e r r o r s and
worsens the forecast performance. Finally, the d y n a m ic forecast errors are larger than the t w o others, especially w h e n the forecast errors are large, emphasizing
the contribution of the e x p l a n a t o r y variable forecast
errors to the forecast error of the inflation forecast.
T h e VAR d y n a m i c forecast fails to improve significantly o v e r the naive n o - c h a n g e forecast in the f u l l
s a m p l e results. Also, the root m e a n squared error of
the VAR is larger than the root m e a n squared error for
the single-equation version that e m p l o y s actual variables as explanatory variables. T h u s , the increase in
the root m e a n squared error m e a s u r e can be attributed
to the forecast errors f r o m forecasting the explanatory
variables. It can be seen more clearly h o w the forecast
errors for inflation increase as the realism of the forecasts in Chart 1 is increased. N o t e that this chart refers
to the m o n e y d e m a n d specification.
T h e shaded area of Chart 1 highlights the forecast err o r s o f t h e t h r e e f o r e c a s t s o v e r the s a m p l e p e r i o d
1984:1 to 1994:3. Over this sample, the forecast errors
for the dynamic model are generally larger than the errors of the other two models. Unfortunately, only the
dynamic model forecast presents a reasonably realistic
test of forecasting accuracy, and over this sample the
VAR f o r e c a s t s relatively poorly. T h i s e x a m p l e illustrates one cost of additional variables in a forecasting
equation—that is, those data series must be forecast into
the future as well in order to forecast inflation. An explanatory variable that is virtually unforecastable m a y
help the statistical fit in-sample but actually hurt the
out-of-sample forecasts. 3 3 Over the 1984:1-1994:3 forecast sample the VAR performs substantially worse than
the n o - c h a n g e f o r e c a s t , o v e r p r e d i c t i n g i n f l a t i o n by
nearly 4 percent during the 1990-91 recession.
Chart 1 presents the forecast errors f r o m static forecasts (forecasts using only actual values), partially dyn a m i c f o r e c a s t s ( f o r e c a s t s that allow only i n f l a t i o n
f o r e c a s t s in f o r e c a s t i n g f u r t h e r in the f o r e c a s t horizon), and d y n a m i c forecasts (forecasts using forecast
values of all variables in the equation). The plots highlight three noticeable features of the forecast error series. First, although the static forecast errors are
clearly the smallest, the static forecast still m a k e s sizable errors, noting that the model errs despite the benefit of forecasting with actual future data not available
in a real-time forecasting exercise. Secondly, the partially d y n a m i c forecast errors are substantially larger
than the static forecast errors, suggesting that the use
of inflation forecast values in subsequent forecasts of
Table 3
Dynamic Forecast Evaluation
Mean Error
Forecast Sample:
Phillips Curve
Money Demand
Phillips Curve
Money Demand
Root Mean
Squared Error
Theil U
1973:1-1994:3
.271
-.492
2.036
2.220
Forecast Sample:
1984:1-1994:3
-.233
-.805
1.347
1.880
22
Econom ic Review
.896
.978
.980
1.368
For the dynamic forecast of the Phillips curve model,
the Phillips curve inflation equation is estimated as well
as r e g r e s s i o n s p e c i f i c a t i o n s f o r the log level of real
G D P and the growth rate of nominal GDP, respectively,
in e q u a t i o n s that do not d i f f e r greatly f r o m a VAR.
From these equations, identities are employed to generate the G D P gap and the nominal G D P growth gap. 3 4
T h e full sample forecasting results for the d y n a m i c
Phillips curve m o d e l are only slightly different f r o m
the Phillips curve m o d e l with actual data for the explanatory variables. Chart 2 shows how the static forecast has m u c h smaller forecast errors than either the
d y n a m i c or partially d y n a m i c forecasts. However, in
contrast to the m o n e y d e m a n d model chart, this chart
shows that the dynamic model forecast errors are not
m u c h larger than the forecast errors of the partially dyn a m i c forecasts. T h e s e results suggest that inflation
forecast errors that b e c o m e incorporated into s u b s e quent f o r e c a s t s of inflation in the partially d y n a m i c
(and dynamic) forecasts have a large influence on inflation forecast errors. T h e r e are relatively small diff e r e n c e s b e t w e e n the f o r e c a s t errors of the partially
d y n a m i c and d y n a m i c f o r e c a s t s , s u g g e s t i n g that the
f o r e c a s t errors f r o m p r e d i c t i o n s of the level of real
G D P and the growth in nominal G D P do not greatly
influence the accuracy of the inflation forecast. In addition, the d y n a m i c forecast has a Theil U statistic of
0 . 8 9 6 f o r the full s a m p l e and of 0 . 9 8 0 for the latter
half of the sample, c o m p a r a b l e to the Theil U statistic
January/February 1995
of 0.855 and 0.940 for the partially dynamic forecast
over those forecast samples. T h e shorter sample period has less variation in the inflation rate. A s a result,
the n o - c h a n g e forecast provides a relatively t o u g h e r
s t a n d a r d f o r the o t h e r m o d e l f o r e c a s t s to i m p r o v e
upon.
Chart 3 presents the d y n a m i c forecasts of both the
Phillips c u r v e and the m o n e y d e m a n d m o d e l s along
with the actual one-year rate of inflation. In the latter
part of the s a m p l e , t h e P h i l l i p s c u r v e m o d e l m o r e
c l o s e l y f o l l o w s the path of i n f l a t i o n t h a n d o e s the
m o n e y d e m a n d model forecast. Neither model appears
to forecast accurately. The t w o m o d e l s often miss the
inflation rate in opposite directions in the early 1980s,
the Phillips curve largely underpredicting the inflation
rate following the serious recession of 1981-83 whereas the m o n e y demand model overpredicts inflation at
that time. Later in the forecast sample, both m o d e l s
appear to err in the s a m e direction.
The Theil U statistics reinforce the apparently better
forecasting of the d y n a m i c Phillips curve model over
the VAR; the Theil U statistics of the dynamic Phillips
curve forecasts are clearly lower than those of the VAR
o v e r the r e s p e c t i v e f o r e c a s t s a m p l e s . 3 5 A l s o , in the
1984:1-1994:3 s a m p l e period, the m e a n error of the
Phillips curve model is m u c h less than that of the VAR,
suggesting a less biased forecast.
The statistical results suggest that the Phillips curve
model m a y be a better forecasting model in real-time
circumstances than the other models presented, including the simple time series model. However, the forecast
e v a l u a t i o n statistics f o r the d y n a m i c P h i l l i p s c u r v e
m o d e l were the best of the m o d e l s e x a m i n e d but w e r e
not m u c h better than the no-change forecast. For exa m p l e , the T h e i l U s t a t i s t i c f o r t h e 1 9 8 4 : 1 - 1 9 9 4 : 3
forecast subsample was close to one, suggesting that
the d y n a m i c Phillips c u r v e f o r e c a s t was not a large
i m p r o v e m e n t over the naive no-change forecast.
T h e VAR includes data series on all the variables in
the Phillips curve model except f o r the potential G D P
m e a s u r e . A d d i n g the potential G D P v a r i a b l e to the
VAR m a k e s the Phillips curve specification a restricted version of the m o r e general VAR m o d e l . Forecasts
of the VAR that include the potential G D P m e a s u r e as
a deterministic variable not only fail to i m p r o v e but in
fact w o r s e n the forecast evaluation statistics for both
forecast samples. This result reflects h o w imposing zero
restrictions on a model (like the VAR) m a y produce improved forecasts. It is unclear, however, whether the resulting f o r e c a s t ( f r o m the Phillips c u r v e m o d e l ) will
be any m o r e useful.
To evaluate f u r t h e r the e f f e c t i v e n e s s of a Phillips
curve inflation forecasting model, the model n e e d s to
be tested in real-time forecasting exercises. 3 6 T h u s , the
Chart 3
Inflation Forecast from Dynamic Models
(Phillips
Curve
versus Money
Demand)
Percent
Federal Reserve Bank of Atlanta
Economic
Review
23
only true test of the m o d e l ' s forecasting ability is to
use it in real-time forecasts and evaluate the results.
Nevertheless, the statistical results in this article suggest that forecasting inflation accurately in real-time
situations with simple economic statistical forecasting
m o d e l s is n o easy task.
T h e results illustrate that the evaluation of forecast
p e r f o r m a n c e c a n c h a n g e s u b s t a n t i a l l y by r e l a x i n g
s o m e of the unrealistic a s s u m p t i o n s i m p o s e d in the
forecast evaluation procedures. This research estimates the simple models imposing the unrealistic ass u m p t i o n s on d a t a a v a i l a b i l i t y , that is, t h o s e u s i n g
a c t u a l v a l u e s on the p r e d i c t o r v a r i a b l e s . D y n a m i c
specifications of the t w o models, each of which generates a completely d y n a m i c out-of-sample forecast, are
then e m p l o y e d . Unfortunately, using the simple m o d els here, the closer forecasting m o d e l s m o v e t o w a r d
approximating the real-time forecasting exercise, the
less i m p r e s s i v e the e m p i r i c a l results f o r the c a u s a l
forecast models. 3 7
Conclusion
This article e x a m i n e s the forecast p e r f o r m a n c e of
t w o simple f r a m e w o r k s f o r m o d e l i n g i n f l a t i o n — t h e
Phillips curve and the m o n e y demand/monetarist m o d e l s — i n a v a r i e t y of s e t t i n g s . T h e o n e - p e r i o d ( o n e quarter) ahead forecasting accuracy of the m o d e l s is
e x a m i n e d first, a n d the results g e n e r a l l y f a v o r both
m a c r o e c o n o m i c m o d e l s relative to the naive " n o
change in the inflation rate f r o m the current inflation
rate" in this forecast horizon. Next, the forecast accuracy of the models is e x a m i n e d for a forecast horizon
of one y e a r into the f u t u r e . In this case, the singlee q u a t i o n m o d e l s e m p l o y " a c t u a l " v a l u e s of the e x planatory variables s o that the relevant c o m p a r a t i v e
statistics do not indicate real-time forecasting accuracy. Then, these simple m a c r o e c o n o m i c specifications
are e x a m i n e d f u r t h e r by generating forecasts f o r all the
variables in each m o d e l for a m o r e truly d y n a m i c fore-
1. In fact, the Social Security example ignores the growth rate
of wages, which determines the growth rate of Social Security revenues. The example abstracts from other variables
24
Econom ic Review
cast. The results of this m o r e realistic exercise show
that forecasting a c c u r a c y of the representative inflation f o r e c a s t i n g m o d e l s d e t e r i o r a t e s (in s o m e w a y s
substantially) and m a y p e r f o r m little better than the
simplest time series forecasting model.
T h e simple e c o n o m i c forecasting m o d e l s e x a m i n e d
in this article d o not forecast inflation effectively. T h e
single-equation m o d e l s , based on the intuition of partial equilibrium m o d e l s that are inherently incomplete,
account for observed relationships in the data rather
than the u n d e r l y i n g f o r c e s that drive i n f l a t i o n . T h e
failings of these forecasting m o d e l s suggest that these
i n t u i t i o n s , in t h e i r s i m p l e f o r m , d o n o t c o n t r i b u t e
greatly to u n d e r s t a n d i n g of the inflation p r o c e s s nor
are they effective for forecasting.
T h e forecasting exercises and the evaluation statistics i m p l y that m a c r o e c o n o m i c inflation f o r e c a s t i n g
m o d e l s s h o u l d be e v a l u a t e d in c i r c u m s t a n c e s that
m o r e c l o s e l y a p p r o x i m a t e the r e a l - t i m e f o r e c a s t i n g
problem. T h e key feature to emphasize, though, about
forecasting inflation is that forecasting equations for
all variables that are in the inflation forecasting model
should be specified. Exercises with static and partially
d y n a m i c forecasts help isolate weaknesses in the inflation forecasting m o d e l s but should not be regarded as
i n d i c a t i v e of e x p e c t e d f o r e c a s t i n g p e r f o r m a n c e b e cause in real-time forecasting there is not the benefit
of k n o w i n g future values of the explanatory variables
as assumed in these t w o exercises.
T h e problem of accurately forecasting inflation remains troublesome, especially in light of the concern
a m o n g p o l i c y m a k e r s and in financial m a r k e t s about
future inflation. D y n a m i c systems of equations appear
to be a m o r e realistic m o d e l i n g f r a m e w o r k for these
inquiries. Fruitful research m a y f o c u s on estimating
d y n a m i c m o d e l s that are by design more structural and
m a y help uncover the sources of inflation. In terms of
optimal inflation forecasting models, research geared
toward d y n a m i c specifications of either a structural or
nonstructural (but restricted) nature will be most useful for real-time applications.
relevant for p l a n n i n g Social Security p a y m e n t s like the
number of new retirees, emergency survivorship benefits,
mortality rate of recipients, and so forth. The net impact of
January/February 1995
the inflation f o r e c a s t error could be higher or lower but
would not affect the point of the example.
2. Turning point analysis tends to concentrate on real (inflationadjusted) output and business cycles, like predicting the next
recession, but some research has aimed similar analysis toward inflation. See Roth (1991) and Klein (1986).
3. Modeling using variables that allegedly cause inflation enables the forecaster to interpret the forecast results as being
due to movements in a particular variable associated with a
set of economic actors (for example, firms, households, and
so forth) included in the model. The interpretability of that
model may be valued by the forecast consumer, as opposed
to forecasts that seem to result from a statistical "black box."
4. The measure squares the forecast errors, thus increasing the
weight of large errors, then sums the squared errors, and
then takes the square root of the sum.
5. Models truly designed for policymaking need more c o m plete structure than the simple forecasting models presented
below. Policymakers would like a fully specified structural
model of the economy to generate forecasts of inflation and
other variables. T h i s model would attempt to m i m i c the
workings of the actual economy in a small-scale, but internally consistent, model that incorporates government policy
actions into the inflation process. The forecasting models
below incorporate only a portion of a theory in the design
of the model and thus offer an incomplete view of the effects of policy on inflation (see Leeper 1993).
6. McNees (1994) provides analysis of the accuracy of official
public sector forecasts (Congressional Budget Office and the
Council of Economic Advisors) as well as private business
forecasts (Blue Chip Consensus Forecasts). He finds the accuracy of both public and private forecasts of CPI-measured
inflation over the past decade to be virtually identical.
7. This statement refers to univariate statistical models.
8. Forecast accuracy criteria may still suggest use of a statistical model; forecast interpretability may be at the cost of less
accuracy. An adequate forecasting model need not be unbiased (that is, have on average a zero mean forecast error).
The chosen model may, for example, have a mean error that
implies overpredicting inflation on average to reduce the
likelihood of realizing higher than anticipated inflation.
Thus, the measure of forecast accuracy can be designed to
fit the criterion most useful to the forecast user.
sults attempt to distinguish between the contribution of "theoretical intuitions" versus other empirical techniques to forecast inflation rather than to determine a "best" model.
12. Frequently, this approach is associated with a Keynesian
view of how inflation is generated. Complete derivations of
the typical Phillips curve models appear in Glassman and
Stockton (1983) and more briefly in Stockton and Glassman (1987) and in Mehra (1988).
13. The following econometric specification is a standard way
to formalize the intuition implicit in the "expectations" augmented Phillips curve relationship (accounting for changes
in the expected rate of inflation by agents' observations of
past inflation rates as seen in the last term):
A
P, = <<> +
~V , ) +
- Ayp) + l"j=[c2+jApH
+ vr
where F is the natural log of nominal G D P , ypt is the natural
log of potential real GDP, yt is the natural log of real (observed) G D P , pt is the natural log of the price level, vf is the
error term, A represents the difference operator, and n is the
number of lagged values of Ap/ ..
The formulation is a slight variation on the simple G D P
gap model. Algebraic manipulation allows the substitution
of nominal G D P into the G D P gap formulation. See Mehra
(1988). Mehra uses a lagged value of the real G D P gap and
the contemporaneous value of the difference between the
nominal G D P growth rate and the rate of growth of potential G D P ; Stockton and Glassman (1987) use a lagged value
of the latter and the contemporaneous value of the former.
14. The limited number of lags of inflation in the estimate is chosen to maximize the degrees of freedom and allow forecast
comparisons over the period including 1973 so that both oil
shocks are in the out-of-sample forecast period. Results of
models with more inflation lags could be noticeably different.
15. As suggested above, it would be reasonable to add other
variables to the specification, but the basic premise of this
article is to examine the contribution to forecasting accuracy of the basic intuition of the respective models.
16. The results in this article employ only the monetary base as
the monetary aggregate; another monetary aggregate would
likely produce different results. Similarly, the Phillips curve
model employs a potential G D P measure produced by Data
Resources, Inc.; another measure of potential G D P would
likely produce different results.
9. Several existing studies examine more extensively the typical macroeconomic models of inflation forecasting. See, for
example, Stockton and Struckmeyer (1989), Stockton and
Glassman (1987), Mehra (1988), and Rcichenstein and Elliot (1987).
17. The inflation forecasting model that results from manipulation of this functional relationship is
10. Three recent articles in the inflation forecasting literature
(Mehra 1988, Stockton and Glassman 1987, and Reichenstein and Elliot 1987) provide useful comparisons of the
forecasting effectiveness of a selection of macroeconomic
inflation m o d e l s for a f o r e c a s t horizon of t w o y e a r s o r
more. The text describes a selective subset of the models
presented in these papers.
where p : is the natural log of the price level, M t is the natural log of the nominal money stock, yt is the natural log of
aggregate output, it is the nominal interest rate, A is the difference operator, and m is the number of lagged values of
the variable.
11. The estimations below avoid ad hoc empirical additions that
may improve the in-sample fit and forecasting statistics for a
model but that are not directly related to that model. The re-
Federal
Reserve Bank of Atlanta
A
P< = So +
+1 »Jmlg2iAyH
+ I7=lg3
AiH + ur
18. An A R I M A O , 1,2) model for inflation forecasting is also
estimated; the results for that model were sufficiently similar to the simple time series model that they have been excluded in the interest of simplicity.
Economic Review
25
19. Estimates of the equations over the full sample are available
from the author.
20. The values of both the Box-Ljung Q statistic (247.8) and
the Durbin-Watson statistic (0.74) indicate autocorrelatcd
errors in the estimated equation. The values of these statistics indicate that the model is poorly specified.
21. The forecast starts from the fourth quarter of 1972 in order
to include data observed in the two oil shocks within the
forecasting sample period. Models are also estimated over a
longer estimation sample (fourth quarter of 1960 to fourth
quarter of 1983), and thus a shortened forecast sample,
starting in the first quarter of 1984.
22. This comparison, however, is only an approximation because
the data used in this study are revised, and real-time forecasting involves forecasting the unrevised values of the data.
23. The article evaluates the forecasting performance of the selected models but is more an exposition on the methods of
forecast evaluation than a rigorous search for the optimal
inflation forecasting model. Financial models of the term
structure have recently been used for inflation forecasting
and have been applied to actual data with some success.
See Fama (1990) for a clear example. The contributions of
these models to inflation forecasting are surveyed in Abken
(1993).
24. The forecast statistics could differ noticeably if the estimates were performed over different forecast periods or
subsamples.
25. The inflation forecasting studies cited above typically use
either eight- or ten-quarter ahead forecasts, leaving relatively few nonoverlapping observations. The longer forecasting
horizon may be warranted, though, in specific applications.
26. Stockton and Glassman (1987) produce forecast evaluations
using forecasts of the explanatory variables as well as the
actual future values of the explanatory variables. Forecasts
using actual data on the explanatory variables had noticeably better forecast evaluation statistics. Such an exercise
highlights the influence on the forecast errors from forecasts of the explanatory variables on the accuracy of the inflation forecasts.
27. It has been argued that Federal Reserve policy became less
tolerant of inflation after 1979.
28. Further inquiry into this issue will be the subject of future
research.
29. Recall that the measure of potential G D P is a number that is
an available estimate, already forecast a number of periods
into the future at the time of the forecast.
30. The specification of the VAR is motivated by the money
demand theory described above. .Clearly, it would be possible to specify a VAR inspired by a Phillips curve perspective. As used here, V A R r e f e r s to the m o n e y d e m a n d
specification.
31. Webb (1984, 1985) has examined extensively the accuracy
of VAR inflation forecasting models. For perhaps a more optimistic appraisal of the accuracy of the technique, see Webb
(1994).
26
Econom ic Review
32. The model illustration below makes this fact more clear:
A M , + lml=lgnAy,-i
Ap, = g 010 +
+
+
+ U
M
W
AM, = g 0 2 0 + Zy = 1 g 2 yAM, + lmj=1g22iAyH
+ lmj=lg24M
Ay, = ¿>030
+
+
+ "2r
+ X m j = x g n £ y H + 1%,^33,-Ag
83ijM*H
A/, = l?040 + lmJ=lg4llAMH
+ X";=1g42,.Ay,_. +
l^g43lAiH
+ Z"j=ig44,Ap, + u4r,
where pt is the natural log of the price level, M, is the natural log of the nominal money stock, y, is the natural log
of aggregate output, /, is the nominal interest rate, A is the
difference operator, m is the number of lagged values of
the variable, and u u is the error term associated with equation /'.
33. In this application, including the relative price of oil as an
explanatory variable increases the root mean squared error
of the inflation forecast.
34. The regression estimates for this model are available upon
request from the author.
= «0 + fi.O'M - V , ) +
+ I%,62ApH
+
+
+ l4j=ig2M-j
+
y
V l
)
vr
AY, = g0 +
y,=y0
- A
_
+ ^
+ l ' „ 7 3 / a ph + e,,
where Yt is the natural log of nominal GDP, 7" is a time
trend, yt is the natural log of real (observed) GDP, /?, is the
natural log of the price level, v,, e,, and
are the error
terms, A represents the difference operator, and n is the
number of lagged values of A/?,_,..
35. The Theil U statistic for the dynamic Phillips curve forecasts is 0.896 for the full sample and 0.978 for the shorter
sample. In contrast, the Theil U statistics for the VAR
(money demand) were 0.980 and 1.368.
36. The potential G D P measure in this model is an exogenous
variable, but it is also known that the data series is estimated ex post. As a result, the series may be revised, and through
the revision process past data observations of the measure
may reflect information not really available at the time period of the observations. The historical data values may reflect important economic information that was not really
available, thereby giving some potential advantage to the
Phillips curve model. This problem is somewhat analogous
to that dealt with in research on the Index of Leading Indicators by Diebold and Rudcbusch (1991).
37. The forecasting exercise ignores the additional real-time
complication of employing unrevised data for economic aggregates like GDP.
January/February 1995
References
Abken, Peter A. "Inflation and the Yield Curve." Federal Res e r v e Bank of A t l a n t a Economic
Review 7 8 ( M a y / J u n e
1993): 13-31.
M e h r a , Y a s h P. " T h e Forecast P e r f o r m a n c e of A l t e r n a t i v e
Models of Inflation." Federal Reserve Bank of Richmond
Economic Review 74 (September/October 1988): 10-18.
Diebold, Fran, and Glen Rudebusch. "Turning Point Prediction
with the Composite Leading Index: An Ex Ante Analysis."
In Leading Economic
Indicators:
New Approaches
and
Forecasting Records, edited by Kajal Lahiri and G e o f f r e y
H. Moore, 231-56. Cambridge: Cambridge University Press,
1991.
Reichenstein, William, and J. Walter Elliot. " A Comparison of
Models of Long-Term Inflation Expectations." Journal of
Monetary Economics 19 (May 1987): 405-25.
Roth, Howard L. "Leading Indicators of Inflation." In Leading
Economic Indicators:
New Approaches
and
Forecasting
Records, edited by Kajal Lahiri and G e o f f r e y H. M o o r e ,
275-301. Cambridge: Cambridge University Press, 1991.
Elliot, J. Walter, and William Reichenstein. " A Monetary Approach to Measuring Long-Run Inflationary Expectations."
Journal of Economics and Business 19 (May 1987): 327-38.
Fama, Eugene F. "Inflation, Output, and M o n e y . " Journal
Business 55, no. 3 (1982): 201-31.
of
. "Term Structure Forecasts of Interest Rates, Inflation,
and Real Returns." Journal of Monetary Economics 25 (January 1990): 59-76.
Glassman, James E., and David J. Stockton. "An Evaluation of
Alternative Price Forecasting Models: Theoretical Considerations." Board of Governors of the Federal Reserve System
Working Paper No. 32, December 1983.
Klein, Philip A . " L e a d i n g Indicators of Inflation in Market
Economies." International Journal of Forecasting 2 (1986):
403-12.
Leeper, Eric M. " T h e Policy Tango: Toward a Holistic View of
Monetary and Fiscal Effects." Federal Reserve Bank of Atlanta Economic Review 78 (July/August 1993): 1-27.
Stockton, David J., and James E. Glassman. "An Evaluation of
the Forecast P e r f o r m a n c e of Alternative Models of Inflat i o n . " Review of Economics
and Statistics
69 ( F e b r u a r y
1987): 108-17.
Stockton, David J., and Charles S. Struckmeyer. "Tests of the
Specification and Predictive Accuracy of Nonnested Models
of Inflation." Review of Economics and Statistics 71 (May
1989): 275-83.
Webb, Roy H. "Vector Autoregressions as a Tool for Forecast
Evaluation." Federal Reserve Bank of Richmond Economic
Review 70 (January/February 1984): 3-11.
. "Toward More Accurate Macroeconomic Forecasts for
Vector A u t o r e g r e s s i o n s . " Federal Reserve Bank of Richmond Economic Review 71 (July/August 1985): 3-11.
. "Forecasts of Inflation from VAR Models." Federal Reserve Bank of Richmond Working Paper 94-8, July 1994.
McNees, Stephen K. "Official and Private Forecasts of the U.S.
Economy." Unpublished manuscript, June 1994.
Federal
Reserve Bank of Atlanta
Economic
Review
27
JReview Essay
An Economist's
Perspective on History:
Thoughts on Institutions,
Institutional Change, and
Economic
Performance
Andrew C. Krikelas
H i s t o r y m a t t e r s . It m a t t e r s n o t j u s t b e c a u s e w e c a n l e a r n f r o m t h e p a s t , but
b e c a u s e the present and the future are c o n n e c t e d to the past by the continuity of
s o c i e t y ' s institutions. T o d a y ' s and t o m o r r o w ' s c h o i c e s are s h a p e d b y the past.
A n d the past can only be m a d e intelligible as a story of institutional evolution.
Integrating institutions into e c o n o m i c theory and e c o n o m i c history is an essential
step in i m p r o v i n g that theory and history.
— D o u g l a s s C . North
W
g
•
g
Institutions, Institutional
Change, and Economic
Performance, by Douglass C.
North (Cambridge:
Cambridge
University Press, 1990), 152
pages, $37.95 ($11.95,
paper).
The reviewer is an economist
in the regional section of
the Atlanta
Fed's
research
department.
28
Econom ic Review
j
1
f o r more than two decades, Douglass North's research has focused
upon institutions and the important role they play in regulating
human social behavior. The ideas that have emerged f r o m this line
of inquiry have led North to develop an innovative f r a m e w o r k for
mJL.
analyzing e c o n o m i c history, presented most c o m p r e h e n s i v e l y in
Institutions,
Institutional
Change, and Economic Performance.
N o r t h acknowledges that this paradigm is not yet fully developed; nevertheless, this
body of research earned him a share of the 1993 Nobel prize for economics. 1
Concerning this award, the Financial Times noted, " T h e Nobel committee's decision to break with tradition and award the prize for work in economic history, rather than economics proper, reflects the growing importance
economists attach to the role of social institutions in providing a f r a m e w o r k
for economic growth." 2 Indeed, the importance of adopting an institutional
January/February 1995
perspective for analyzing economic events has grown
in recent years in the w a k e of the collapse of the socialist economies in Eastern E u r o p e and the Soviet Union.
W h i l e m a n y neoclassically trained economists h a v e
been asked to counsel the leaders of nations m a k i n g
the transition f r o m centrally p l a n n e d to f r e e m a r k e t
economies, m u c h of their advice has proved inadequate
if not unhelpful. The main reason for this inadequacy is
that most neoclassical economic models are based upon two assumptions: that society's institutional infrastructure is essentially free-market oriented and that it is
unchanging. These m o d e l s , therefore, are not suitable
for analyzing situations in which the institutional infrastructure diverges m u c h f r o m the free-market stand a r d or t h o s e in w h i c h i n s t i t u t i o n a l c h a n g e is t h e
order of the day, and they are particularly inadequate
for those in which both conditions prevail simultaneously, as is the case currently in Eastern E u r o p e and
the countries of the f o r m e r Soviet Union. B e c a u s e institutional economists have not yet developed an alternative behavioral model that accounts for the process
of institutional change and its related economic conseq u e n c e s , they too h a v e not been able to o f f e r m u c h
specific advice except to criticize the calls for shock
therapy m a d e by a n u m b e r of mainstream economists.
Although m a n y economists remain skeptical that an
adequate alternative to the neoclassical paradigm will
ever be d e v e l o p e d , I would argue that N o r t h has already m a d e substantial progress toward this goal. The
g a m e theoretic model he describes in Institutions,
Institutional Change, and Economic Performance
is both
rigorous and intuitive and deserves a m u c h wider reading a m o n g social scientists than it has received to date.
T h e purpose of this review^therefore, is threefold: to
stress the b o o k ' s importance, to explain the essential
features of North's model, and to demonstrate the model's usefulness as an analytical tool.
iVorth's View of History
F r o m the outset, it is important to r e c o g n i z e that
what North actually offers is a coherent and detailed
theory of social history. His methodology allows one to
c o n f r o n t , and a n s w e r directly, t w o related questions:
W h a t is history, and why does it unfold as it does? In
c h a p t e r o n e N o r t h i n t r o d u c e s a t e a m sport a n a l o g y ,
used throughout the book, that illustrates his theory: Institutions, he states, "are the f r a m e w o r k within which
human interaction takes place. T h e y are perfectly analogous to the rules of the g a m e in a competitive team
Federal Reserve Bank of Atlanta
sport. That is, they consist of formal written rules as
well as typically unwritten c o d e s of conduct that underlie and supplement formal rules, such as not deliberately injuring a key player on the opposing team. A n d
as this a n a l o g y would imply, the rules and i n f o r m a l
codes are sometimes violated and punishment is enacted. Therefore, an essential part of the functioning of institutions is the costliness of ascertaining violations and
the severity of p u n i s h m e n t " (4).
From North's perspective history can be characterized as the r e c o r d of an e v o l v i n g g a m e . W h i l e the
g a m e is complex in its detail, its f u n d a m e n t a l structure
is relatively simple, consisting of just three primary elements: institutions, organizations, and individuals
(rules, teams, and players). Although most economists
would be inclined to describe the g a m e as an individual
sport, North assumes it to be a team competition. H e is
not saying that individuals d o not play an important
role in the game. W h a t he does imply is that there are
m a n y social, political, and e c o n o m i c o b j e c t i v e s that
can be obtained by individuals only through team effort. In order for each of the players to succeed as ind i v i d u a l s , t h e y m u s t be p r e p a r e d t o b e h a v e b o t h
cooperatively, in order to produce social wealth, and
c o m p e t i t i v e l y , in o r d e r to c o n s u m e it. S t r u c t u r a l l y ,
then, the g a m e that North describes could be viewed as
a cooperative social competition.
At stake in this competition, according to North, are
the substantial gains f r o m trade that can be produced
by an economic system organized around the dual principles of specialized production and market exchange. 3
These returns, in fact, are exactly the same profit opportunities that fuel the competitive desires of the actors in neoclassical e c o n o m i c m o d e l s . In contrast to
m o s t m a i n s t r e a m e c o n o m i s t s , w h o generally a s s u m e
that players always obey the mies, thereby producing
none of the social costs associated with cheating and
other antisocial behaviors, North assumes that cheating
is inevitable and that referees, therefore, must be hired
to resolve conflicts. T h u s , in addition to the role played
by teams engaged in pure e c o n o m i c activity, N o r t h ' s
model contains an explicit role for social and political
teams as providers of the services required to maintain
orderly economic competition.
The services that social and political teams provide
w i t h i n the c o n t e x t of the g a m e can be d i v i d e d into
three distinct categories (see Table 1). The first set of
teams, type I social and political teams, are those that
act as referees; their m a i n p u r p o s e is to detect rulebreakers and to resolve conflict. Such refereeing services are provided both formally by the state through
the police force and judicial system and informally by
Economic
Review
29
Table 1
The Four Types of Organizations That
Participate in Cooperative Social Competition
Type of Organization
Objective
Economic Teams
To produce and distribute
economic goods and services.
These teams are the focal point
of the competition.
Type 1 Social and
Political Teams
To detect and punish teams and
players who violate established
rules and to resolve conflicts.
Type II Social and
Political Teams
To teach players the rules.
Type III Social and
Political Teams
To lobby for changes in the
existing rules.
parents within families, clerics within congregations,
and managers within business offices. Although these
monitoring services are costly to provide and only imperfectly achieve their goals, they nevertheless play an
i n d i s p e n s a b l e role in support of the u n d e r l y i n g economic competition.
A second and related set of teams, type II social and
political teams, have as their objective the socialization
of the individual—the process by which he or she is
both taught and persuaded to play by the established
rules of the game. Unlike most economists, w h o would
assume that players are born with a c o m p l e t e understanding of the rules, North contends that individuals
b e c o m e aware of the rules only through a lifelong educational process. 4 This process begins within the family with the lessons taught by p a r e n t s , siblings, and
other relatives. O v e r time, however, the socialization of
the individual e x p a n d s to i n c l u d e lessons taught by
m a n y other teams and their players, including educators in schools, clerics in congregations, employers in
businesses, and even politicians as leaders of political
parties. Because the process by which each person obtains his or her education is unique, it is unlikely that
t w o individuals will have identical perceptions of any
given social situation. As a result, North allows for social conflicts to arise not only f r o m willful cheating on
the part of teams and players but also f r o m differences
in the socialized expectations of the parties engaged in
a particular exchange.
30
Econom ic Review
T h e third set of social and political teams participating in this c o m p e t i t i o n is quite different f r o m the
first two. W h i l e type I and II teams serve to e n f o r c e
and perpetuate the existing g a m e rules, type III teams
seek to rewrite these rules. Such institutional c h a n g e s
m a y be introduced cooperatively, through prescribed
political and legislative channels, or uncooperatively,
through the use of f o r c e in w a r or revolution. In each
case, however, the social and political teams that init i a t e t h e s e r e f o r m s d o s o o n b e h a l f of e c o n o m i c
t e a m s — t e a m s w h o s e competitive interests would be
served by c h a n g e s in specific rules. Although the m o tivations f o r such changes could stem f r o m m a n y
sources, t w o of the m o r e likely would be a substantial
shift in relative p r i c e s that finally allows a t e a m to
hire p r e v i o u s l y u n a f f o r d a b l e l o b b y i n g s e r v i c e s a n d
the a d v e n t of an i n n o v a t i v e t e c h n o l o g y w h o s e e f f i cient use conflicts with existing rules. In other w o r d s ,
in addition to allowing f o r the possibility of institutional change and social evolution, N o r t h ' s model
identifies the origin to be the entrepreneurial b e h a v i o r
of this g r o u p of social and political t e a m s and their
players.
Although North does not take the team sport analogy quite as f a r as this essay has, this elaboration fits
well within North's view of history as the record of an
e v o l v i n g g a m e . U n l i k e in m o s t sports, h o w e v e r , the
rules of this g a m e are not fixed but are subject to constant change. E a c h successive rules change forces all
teams and players to reevaluate and modify their corporate and personal strategies, with the result that even
small institutional changes m a y ultimately have a prof o u n d impact on the o u t c o m e of the competition. In
order to understand the outcomes of any given cooperative social c o m p e t i t i o n , it is n e c e s s a r y to collect as
m u c h of the f o l l o w i n g information as possible: (1) a
complete set of the rules of the g a m e at the beginning of
the competition; (2) a list of the relevant c h a n g e s to
those rules over time; (3) a list of the four types of organizations c o m p e t i n g over time, as well as a sense of
their corporate strategies; and (4) a list of the individual
players, the portfolio of team affiliations they maintain,
and the personal strategies they adopt under each set of
prevailing rules.
Applying the Model
North's model can be used for a variety of analytical
purposes, one of which, of course, is interpreting the
events of a specific historical period. In fact, North de-
January/February 1995
votes several chapters of his book to just such an investigation, focusing specifically on the economic history
of the Western Hemisphere. In a world in which information and capital flow freely and transactions costs
are negligible, neoclassical e c o n o m i c theory predicts
that economic convergence will occur, especially in the
long run. F r o m this perspective, the divergent e c o n o m ic p e r f o r m a n c e s during the last t w o hundred years of
the U n i t e d S t a t e s and C a n a d a on the o n e h a n d and
most of Latin A m e r i c a on the other provide a persistent
puzzle.
North solves this puzzle by pointing out that, in a
world in which institutions shape the behavior of the
teams and players engaged in cooperative social c o m petition, c o n v e r g e n c e d o e s not necessarily occur. Alt h o u g h t w o s o c i e t i e s m i g h t be i d e n t i c a l l y e n d o w e d
w i t h l a n d , l a b o r , c a p i t a l , a n d o t h e r p r o d u c t i v e resources, if their institutional structures are significantly different then they are likely to achieve dramatically
different e c o n o m i c results, even in the long run.
For example, the institutions of private property and
individual rights that North Americans inherited from
their British colonizers established powerful behavioral
incentives f o r private a c c u m u l a t i o n of both physical
and h u m a n capital, decentralization of e c o n o m i c and
political decision making, and attention to overall productive efficiency. By contrast, the state property rights
and central authority that the Spanish c o n q u i s t a d o r s
i m p o s e d on colonial Latin A m e r i c a p r o m o t e d a m o n o p o l y o v e r both e c o n o m i c and political r e s o u r c e s ,
encouraging redistributive, rather than productive, activities. North argues that the divergent economic performances observed in the Western Hemisphere during
the last t w o h u n d r e d y e a r s reflect the very d i f f e r e n t
ways in w h i c h these cooperative social c o m p e t i t i o n s
have been organized and conducted.
bridging the Gaps between
Academic Disciplines
In addition to being useful for analyzing the history
of a particular society, N o r t h ' s g a m e theoretic f r a m e w o r k has the potential to serve as a m e t h o d o l o g i c a l
bridge between the diverse branches of the social scie n c e s and h u m a n i t i e s . T h e social s c i e n c e s — s u c h as
economics, political science, sociology, and social psychology—deal with the institutions and functioning of
h u m a n society and the interrelationship of its m e m b e r s .
History, although s o m e t i m e s c a t e g o r i z e d as a social
science, is generally considered one of the humanities,
Federal Reserve Bank of Atlanta
which can be defined as the branches of learning concerned with human culture and values and which usually include languages, literature, history, mathematics,
and p h i l o s o p h y . A l t h o u g h s c h o l a r s in e a c h of t h e s e
fields conduct research that would benefit greatly f r o m
interdisciplinary discussion, relatively little such comm u n i c a t i o n h a s o c c u r r e d b e c a u s e there h a s b e e n n o
unifying theory to identify the ways in which these disciplines are related.
N o r t h ' s g a m e theoretic f r a m e w o r k g o e s a long w a y
t o w a r d p r o v i d i n g j u s t such a u n i f i e d p e r s p e c t i v e . It
serves as an intuitive lens through which various dis-
North's game theoretic framework has the
potential to serve as a methodological
bridge between the diverse branches of the
social sciences and humanities.
ciplines c a n be viewed as one or m o r e aspects of coo p e r a t i v e social c o m p e t i t i o n . History, f o r e x a m p l e ,
can be thought of as the record of all o u t c o m e s prod u c e d by a given social c o m p e t i t i o n , with other hum a n i t i e s f o c u s i n g s p e c i f i c a l l y o n s u b s e t s of t h o s e
o u t c o m e s . Within this f r a m e w o r k , items of art and literature would be characterized as cultural if they support the socialization e f f o r t s of the type I and type II
social and political t e a m s and countercultural if they
p r o m o t e the institutional c h a n g e o b j e c t i v e s of type III
teams.
Social scientists, on the other hand, in addition to
studying the o u t c o m e s of cooperative social competition, also seek to understand the underlying processes
that g e n e r a t e these results. W h i l e m a n y e c o n o m i s t s
m a y believe that theirs is the only g a m e in town, North
a r g u e s q u i t e c o n v i n c i n g l y that s o c i a l a n d p o l i t i c a l
teams play an important and integral role in cooperative social c o m p e t i t i o n . In f a c t , N o r t h ' s f r a m e w o r k
clearly suggests that greater interdisciplinary c o m m u nication a m o n g economists, political scientists, sociologists, and social psychologists is required if a better
model of cooperative social competition is to be developed.
Economic
Review
31
Conclusion
In order to m a k e sense of history, it is essential to
have a model that adequately accounts for the role institutions play in determining social, political, and econ o m i c events. N o r t h ' s p a r a d i g m is o n e such m o d e l .
Although not yet fully specified in terms of the formal
m a t h e m a t i c s t h a t h a v e b e c o m e the s t a n d a r d of the
m a i n s t r e a m e c o n o m i c s literature, N o r t h ' s g a m e theoretic f r a m e w o r k nevertheless represents an important
step toward a better understanding of why history unfolds as it does.
Because North's behavioral assumptions diverge
s u b s t a n t i a l l y f r o m those c u r r e n t l y a d o p t e d by m o s t
economists, it is likely to take some time before this new
p a r a d i g m b e c o m e s a c c e p t e d in the m a i n s t r e a m e c o nomics literature. Conversely, because North's methodo l o g i c a l a p p r o a c h is e s s e n t i a l l y that of n e o c l a s s i c a l
economics, it has been criticized by institutionalists and
other noneconomists who are put off by its relative formality.
N o r t h h i m s e l f c o n s i d e r s his m o d e l an e x t e n s i o n ,
rather than an a b a n d o n m e n t , of traditional e c o n o m i c
analysis: " I n t e g r a t i n g institutional analysis into econ o m i c s and e c o n o m i c history is redirecting emphasis,
but not abandoning the theoretical tools already developed. Redirecting the emphasis entails m o d i f y i n g the
n o t i o n and i m p l i c a t i o n s of rationality, i n c o r p o r a t i n g
ideas and ideologies into our analysis, explicitly studying the costs of transacting for the functioning of political and e c o n o m i c markets, and understanding the
c o n s e q u e n c e s of path d e p e n d e n c e f o r the historical
evolution of e c o n o m i e s " (135).
N o r t h ' s n e w i n s t i t u t i o n a l f r a m e w o r k o f f e r s real
p r o m i s e as a m e a n s by which to facilitate interdisciplinary communication among economists, institutionalists, and other social scientists. Institutions,
Institutional
Change, and Economic Performance should become required reading for scholars in all of these fields.
Notes
1. In a 1992 article, North says of this paradigm: "All of this,
therefore, does not add up to anything as elegant as a theory. A
dynamic theory of economic change is the objective; but what
we have so far is a set of definitions and principles and a structure that make up some of the essential scaffolding necessary
to a theory of institutional change" ("Institutions and Economic Theory," American Economist 36 [Spring 1992]: 6).
2. "Economic Historians Win Nobel," Financial
ber 13, 1993, 7.
Times,
Octo-
3. Some form of market exchange occurs in nearly all economies that engage in specialized production. This is not to say
that the market structures in centrally planned and free-market
economies are not qualitatively different. However, in both of
these economic systems the goods produced by specialized
Econom ic Review
32
teams are redistributed to other teams and players within society through some form of market exchange.
4. It is at the level of the individual that North's behavioral model diverges most substantially from the neoclassical economic
paradigm. While many economists would assume that individuals have the capacity to know and understand completely
the rules governing cooperative social competition, North
subscribes to Herbert A. Simon's view of bounded rationality,
which assumes the individual has only a limited understanding of these formal and informal rules. For a recent discussion
of bounded rationality, see chapter 6 of James G. March and
Herbert A. Simon, Organizations (Cambridge, Mass.: Blackwell, 1993).
January/February 1995
• y
,
-
.
.
/ '
, , ¡
: ^
„ ;
'
'I
,
-•
.
"
;
- . , -
.
;
'
:
V v
v.
\
• ^
>
v
;
v
v .
"
?
' V
;
:
^
-
• • -
,
-
^
. . .
-
:•-.••"
. . . .
.
• -x ,
./
'
V cj-
-
,
;
-
'
'
,
.
-
.
i
f
...
-
<>
-
'
: . 's
-
-1
• VJ
. Í
:
•
i
-
.
-
-
.
•
'
•N
Digitized: for FRASER - /
-
-
, •• .
—y
i
'
,.
. . . .
V
-
'
•
-
',. •
..
-
;
.
•
/
,
•
- '
^
-,
,
1
• -
-
-
-
....
. .(
.
/
r .
;
Bulk Rate
U.S. Postage
PAID
Atlanta, GA
Permit 292
BANK OF
Public Affairs
Department
104 Marietta Street,
Atlanta, Georgia
( 4 0 4 )
N.W.
30303-2713
5 2 1 - 8 0 2 0
®
printed on recycled paper
@FedFRASER